Click here to make a new post
for Hardware support for SolarisX86

HI!

Just though I'd let you know that 2940 support is now available as a patch: 107573-01 SunOS 5.7_x86: cadp: Add Ultra2 SCSI support As it's a new driver, it may be only in the contract area, but at least it's out there.

Not having U2W Adaptec drivers caused major headaches in putting in the next 126GB of RAID on an NSF server. This year, we had to use Solaris because of compatibility and reliability. Next year, we will look at Linux.

Linux, Apache, SendMail, etc. do better than the commercial products because they are developed for customers by customers. They avoid all of the overhead of marketing and sales, all of the problems of interactions between these groups and engineering, scheduling, ...

IBM, Sun and Apple all make OSs and development tools to support their own proprietary hardware. They are practically the lsst companies standing with this strategy.

The PC model is winning precisely because it decouples so many of the decisions, allows independent actions of HW and SW vendors, etc.

Solaris on Intel is Sun's opportynity to win in this market and skim some of the top-of-the-line reenue that Windows NT now garners.

Instead, Sun apparently views Solaris on Intel as competition to their workstations. Maybe, but ignoring the problem and under-funding Solaris/Intel development merely means that Windows NT and Linux will eventually take over the Solaris market.

Emphasize reliability, support it better than Microsoft or Linux, keep it up to date with the latest hardware, and maybe Sun won't be pushed into the upper 0.05% of the computer server market. (Still a big market, but not where I would want to be. Compaq will eventually buy Sun, given Sun's current strategy.)

Lew

I understand that more 80,000 copies of Solarisx86 were given away. Now let's say a few of those copies actually produces a viable product. What can they do with it? How is sun supposed to support the end customer of such a thing when they barely support the developer. How can they sell the product in a small scale environment?

What about the big boys like NCR and Seimens, what are the support mechanisms they use? What kind of boxes do they use? Can they buy video cards which are End-of-Lifed? I can just see them trying to get a platform ready let's say using XYZ-RAID controller, 6 months QA and development. Then they are ready to ship boxes and *POOF* they can't buy that controller anymore.

What are these big vendors going to do for service. A 14 month old controller died it has been EOLed, the driver is not available for the next closest (most cost effective) model for the customer. Without shopping in the used closet. NCR isn't going wait for 6 months to fix that box...while the next DU/MU comes out...what will they do?

I understand not supporting Joe's-Garage-Video Cards but this is a fast market. What is needed is a separate support unit for SolarisX86, It must have a different structure than Sparc because it is customer driven, which is market driven, Hardware is disposable not replaceable. Many of the players are in the audience. You got a snot kicking OS here, so Use it, Sell it, Make a separate X86 support site run off an Intel box. One of the reasons you can't market it is because you don't use it. Let's see it. Man you guys are missing the boat here.

What does a GM of Seimens think when he reads that article that Linux is M$'s only threat; the article goes on to mention all the other UNIX on X86 OS's including Zenix but never mentions Solaris. Get somebody reading all the rags, and make sure that Solarisx86 is in the next article that uses Linux, SCO, FreeBSD as a Reference. Even if it performs lower it is there. Give Ziff-Davis a donation (facilitation payment). I bet GBit NIC on Solarisx86 will whitewash a NT box on the IDENTICAL (READ DUAL BOOT) Machine. How many drivers are available for Gbit Ethernet.

When penguin can't scale Solaris can. So capitalize on this, be ready to deliver Solarisx86 where Linux falls short. It will be a narrow window all you have to do is lift the shades.

Like it or not Superior or not Intel hardware is here to stay. Maybe you guys could even influence what makes it to market. Wouldn't that be a a gas :-).

Give Visual Workshop C/C++ away for FREE for non-commercial usage to go along with the FREE Solaris x86 give way.

Sun's entire reason for giving away Solaris x86 is an attempt to get developers writing code for the Solaris platform (mainly Sparc). This is the reason SCO starting giving UnixWare away for FREE long before Sun's Solaris promotion.

Solaris x86 gives no advantage over UnixWare or Linux from a developers point of view. Only when bundled with Visual Workshop does Solaris Intel become the obvious UNIX development platform. Even Linux people would have to admit this.

It costs Sun nothing,as Sun is not exactly getting rich selling Solaris x86 Visual Workship licenses.

The license would be for one user to prevent the Intel box from being used as a server.

Does Sun seriously think anyone is going to pay $3000 for Sun Workshop for Intel when they could get Windows NT and VC++ 6.0 for 1/10th the price?

And when I say Visual Workshop, I don't mean Java I mean C and C++ Workshop. In the real world people use other languages besides Java and real Java developers don't use Visual Workshop.

Imagine year after year college students entering the workforce already knowing Solaris and Visual Workshop since thats what they did their homework assignments on.

Bryan Althaus Bear Stearns, & Co. Inc.

As with many of the posts to this web page my wish for Solaris x86 is wider ranging than just the hardware options listed above.

My department has been running Solaris x86 for approximately 3 years now. Originally our choice for Solaris x86 was based upon 2 things: price/performance when compared with the sparc hardware available to us at that point, ease of integration and support into our existing Solaris/sparc based environment. New issues have arisen which have forced us in to a dual boot environment, thus even with the advent of the Darwin series, leaves us still in the market for x86 based solutions.

We have been extremely pleased with Solaris x86 in our academic environment. Currently about 1/4 our equipment is x86 based including two multiprocessor PPro based machines and our most utilized teaching lab.

My concern lies with support for current hardware shipping in off the shelf PC's today. We do not use specialized hardware in our environment but we do need support for the common AGP cards, UltraATA drives, sound cards, etc. coming in PC's being built today. We will be purchasing 25 or so new PCs to update an aging lab and our greatest concern currently is that the video board will not be supported.

It concerns me that I'm not completely aware of what replaced the old Driver Update system used previously in 2.5/2.6. It seems that the MU and quarterly releases has replaced this but how does this affect the home user?? My school has a service contract but there are many students taking advantage of the promotional offers, where do they get the updated driver and new driver support? This seems to be a common concern currently on the list.

Outside of work, I would like to see better support for laptops. I am in the market for a new laptop for personal use and would like very much to run Solaris on it, however I have been greatly discouraged by the constant posts to the x86 mailing list regarding lack of support for PCMCIA and video chipsets in newer laptops.

Although I would still prefer to purchase and support Sun hardware (we've had tremendous reliability and longevity from our Sun hardware), our needs for x86 based solutions will continue. Thus I would like to see Sun continue to support, expand, and improve their Solaris x86 product which I think is an excellent product.

Thanks for the opportunity to share opinions and concerns.

-- John

Sun/Scott

I've developed and sold solutions on Solaris/SunOS for over 10 years. I've been looking at Solaris x86 since 1993. X86 has improved, but I still get nervous recommending it for fear that it won't run on a "standard" PC without some unprofitable, hard work.

One current hot business for us is scalable turn-key Java-based application/Web/file server solutions. Solaris is a great platform but NT eats the bottom end of this market. Solaris x86 could be positioned more visibly a la "Cobalt Qube" with specific PC vendors pre-installing (eg. for ISPs etc). We could sell lots of these. Drop the x86 price to get some volume and geek loyalty, and push the marketing on the scalability from x86 to "real" SPARC servers. Open the driver source, set up driver developer forums, and support the major card manufacturers (eg. Adaptec U2W SCSI - I want it NOW!) when they release new product. Perhaps increased visibility and volume will encourage them to develop drivers.

Thanks for soliciting some feedback. --Richard.

If you are serious about the PC, then handle it seriously: you have to keep up with the hardware so everyone can have their favorite UNIX on their latest toy. No OS is cool enough to make one proud of their circa 1994 P120 notebook.

But that's just one kind of Solx86 user. I'm sure somewhere, there's someone using x86 solaris in a vertical application. Perhaps controlling a remote telescope or something. Due to lack of 3rd party and even lack of Sun support for x86 Solaris, the open source community is really the only other people using it, or considering using it, and the current state of things makes that one a hard sell too, even if x86 Solaris was free - Linux is still better at supporting the latest gadgets. Granted, you don't need a PC Card or audio support on a PC Apache/Solaris web server, but you should at least be able to use the latest SCSI adapters from the major vendors, and the latest PCI ethernet. Or the latest native PC drives, for that matter. If you can't support at least these basics, then forget it. Don't bother. Can the product. Microsoft gets everybody else to write drivers for them People fall all over themselves to get the "Made for Win95" logo. Buslogic/Mylex are really the only vendors I have ever seen that go out of their way to support Solaris x86 (and their boxes are the only place where I have seen a "Solaris x86 OK" logo). Coincidentally, the Linux support from BusLogic is also terrific. I understand that BusLogic is very forthcoming with developer information to support their products. Maybe you could take a clue from them. You don't have put Solaris on the GPL, but you ought to make it freely redistributable, and easy and cheap to develop for.. Face it, UNIX on PC users are cheap bastards! But for good reason - a lot of them are students, or hobbyists, or entrepreneurs just starting on a new project or idea with working capital out of their own pocket. I think there's a much larger potential for x86 Solaris as an ambassador for Sun than as a stand alone money making division of Sunsoft. One of those ideas may pan out to be something requiring heavy lifting servers - who will they call? So my suggestions are, (1) make X86 free for download (or purchase on CDROM a la RedHat), along with the SDK, (2) to make the SDK complete and useable for anyone with the time and inclincation to develop drivers for whatever piece of hardware they have. You might even consider allowing 3rd parties to re-package and redistribute it, as is allowable for Linux, as long as they do it GPL-style. It's OK by me if you want to stipulate that distribution of those drivers for profit requires purchase of the full x86 and SDK license (and yada yada...) and use of the free Solaris is restricted the same way (and yada yada). Basically, just work it out so that developing on x86 is as cheap and easy (read: free) as possible - as cheap and easy as it is for Linux. Do that, and you stand a chance of Solaris x86 sticking around (if Sun wants it to stick around). I'd even be happy with a X-server-less Solaris x86, as long as it was easy to add Xfree86 to it, or if (a recent version) was bundled with it. Do that, and Sun only has to support text mode, and virtual consoles, not the whole rack of new video cards... Spend those resources getting SCSI, disk, and support for the new buses (USB, PC Card) and motherboards up to speed.

That's all I have to say. Most of it was probably wishful thinking. :)

These are general comments on Solaris X86. Firstly I would like to say I have some appreciation of the economics of this situation in that sun does not derive a terrible lot of revenue from solaris x86. However this can be a chicken and the egg issue. We don't use solaris X86 terribly much because of (1) lack of support from sun mainly in the device driver area - mostly video and notebook support and (2) lack of 3rd party application support. I contend that reason 2 follows because of reason 1. It seems to me that solaris x86 has a lot of competition from free unix's like linux, freebsd, netbsd, openbsd etc. I think sco unix also has existing market share that could be hard to grab. I think solaris is superior to sco unix and the other free unixs. I think the free unixs do some things better, especially driver support, although nobody really does notebook support very well. This could be a niche solaris x86 could go after, although the market is small compared to desktops and the hardware is enormously varied. I guess the obvious thing to do would be to go after the notebook market leaders first and see how much extra effort is required after that to expand the notebook market to other vendors. This last comment applies equally to other device drivers like video cards - go after the market leaders first. In the video area there seem to be virtually no 3D cards supported these days. There is no opengl support which would be absolutely crucial to some commercial apps like CAD. I say pick some decent 3D cards, provide good device drivers and openGL support and lean on some of the CAD vendors.

The best thing about Solaris x86 for me is that in terms of supporting another platform, it is not an entire platform, if you already have solaris on sparc as we do. We have solaris and irix which I count as 2.3 platforms because solaris means solaris on intel and sparc. If we had linux instead of solaris x86 I would count 3 platforms. Finally it seems that solaris on x86 could be more strategic to the future of Sun than Solaris on sparc. It seems the sparc market is not that big and not really growing much. I don't know if merced will be the next wonder chip to grab lots of market share but lots of companies think so. So while it is important to have a foot in both camps I think a little more weight should be placed on the Solaris x86 foot.

We tend to like a one vendor HW and SW solution for our servers something like solaris on sun sparc boxes. For our desktops this is less critical, although some of the contracts we work on may be classed as critical and may require a one vendor solution.

Finally the latest problem we have with solaris x86 is the lack of support for EIDE drives > 8GB.

I would like to see support for audio drivers like Yamaha. I would also like to get better drivers for pcmcia.

My message to Sun:

Tell us what you're thinking!

It's obvious that there are at least three major problems with Solaris/x86/Intel:

1) device driver support is lagging;
2) support for laptops is lagging;
3) application (incl. Java) support is lagging

These are obvious to anyone who follows Sol/x86 for any length of time so we can't be telling you anything new. So why not be unGates-ian and tell us exactly what your business plans are wrt Sol/x86? Instead of us telling you what we all already know why don't you just put us all on the same page? We're tired of blank pages.

Why do we never see any updates to the sound blaster drivers. Its pretty bad when I have to go hack /platform/i86pc/boot/solaris/driverdb to add the Plug and Play ID for my sound blaster card just to get 2.6 to support my sound card in half duplex mode. No support for the PCI cards is even mentioned.

It sure would be nice ...

This one seems to be older than the Solaris x86 video card support. Solaris started support for the UW SCSI when the U2W came out, at least on Sun hardware. U2W has been out for some time and still there is no support for it by Solaris. I understand you have to wait and see what technology is going to make and is not but how long do we have to wait. This is not support for some off the wall technology but for reliable and performance technology. SCSI is not going away, UW SCSI is becoming default and U2W is just arround the corner for the mainstream market.

Thank you Sun for your support so far and keep up the good work.

Thank you.

When will we see support for this card? In the PC world hardware turns about every six months. By the time we get driver support for hardware from Sun its time for new hardware. I'm not suggesting you support every single piece of hardware that comes out, but support for the most popular and perhaps the most stable/powerfull hardware would be nice. The Riva TNT 2 is coming out and we do not have drivers for the original Riva TNT. I've been running Solaris on a Sparc 20 for some time now and I love the multiple monitor support taht Sollaris offers. NT does not support multiple monitors at the OS level. There are a few video cards from different vendors that support multiple monitor at the hardware level, very few and those cards are not at the top anymore.

Solaris x86 would be a easy replacement for NT but it needs to support more hardware, again, not everything but at least support the best from the latest hardware.

Thank you. Keep up the good work and keep that stock price going up.

I just want to see the colors!!!

Hi!

Many thoughts of mine have been echoed by other people here, but more useful are the real world examples. So I'll keep my contribution short.

Went to sign up for the Sun Developer program. All the graphics sutff is for SPARC, or Win32. This doesn't matter, because ....

Went to buy a new video card. Can't get a driver. That's OK, I signed up for a developer program ....

but my registration was misplaced. Oh well, I'll have to make more money so I try to ....

...sell some systems, but the software vendor don't do Solx86 and the O/S doesn't provide support for the SCSI card. Once again, ....

tried to write a bit of code, can't understand the documentation. Eventually, realiased I needed a patch set so I ...

downloaded an update for Sol7, had to get the 2.6 patches as well (blew the d/l out by some 500%!).

---

Saw a Sun Demo, cried - 'coz he was using a laptop.

---

I love the OS. Came to it after the demise of Amiga.

Supporting even one or two laptops is really important.

There are decent GFX cards out there, and the AGP hasn't even rated a mention yet. No Open GL or Java3D.

Despite the old Amiga's drawbacks, it could do amazing graphics with half the power of todays beasties. Surely this is not too much to ask. Pete.

Hi, This isn't a rant but a question. I have successfully, after a couple of attempts, installed the Solaris 5.7 X86 product onto one of my systems which dual triple boots, NT-40 Server, W98 and RedHat Linux 5.2.

The only problem I seem to have is that in comparison with the other OS'es, the Solaris 5.7 X86 product needs a lot more memory. So my question is thus... why is so much memory needed for an OS to perform???

When I initially installed the product I only had about 32 meg of RAM, two 16 meg SIMMS. After the installation everything was very very slow. Since then I have increased the total memory on the system using a DIMM to give me a total of 100 meg. The performance did improve greatly but it is still slower than the others OS'es running under the same hardware.

!!???What gives????

C. J. Parker, An interested Computer Professional.

Briefly: -I've arrived to manage a shop with 2 RS/6000, 1 Ultra 450, and 6 NT Dell & Compaq boxes. -

I received the above notebook, and would've liked to run Solaris on it, but have put up Linux because of 1) hardware support and 2) ease of dual boot with NT. I'd be happier with Solaris and this would have become a talking point in Sun's favor in the office.

-We need to put up a monitoring/logging box. This kind of application is perfect for Intel hardware. This will be a Solaris 2.7 x86 box. x86 hardware has a wide range of applications for Solaris; however, it needs better support if its going to get serious money spent on it by people whose reputation is on the line. --

I urge Sun to give some thought to stationing some engineers at a few hardware companies (Compaq for servers, e.g. their rackmounts, Dell for better, e.g. Optiplex, desktops, and Toshiba for laptops) as a start to provide driver support for standard (the reverse of what Intel has done at Be; there hardware is supporting software, but the idea is similar). This type of worldwide full-range support allows Sun's product to become more useful and hence ubiquitous in the office; this larger installed base drives more sales of servers and software.

Please add a driver for this very common chipset.

As a long-time professional developer, I find Solaris to be the most "Developer-Friendly". I develop for lots of Unix systems (including AIX, HPUX, Irix, and several others), but the ability to work from home, on Solaris x86 makes it extremely attractive. I think Sun has done a hell of a job when it released x86, as it's almost the same as working on a Sparc... Well, almost !

I'm not saying almost for the differences between Intel's and Sparc superior CPUs, I'm saying almost for what I find to be very frustrating that x86 hardware support on Solaris lags behind the industry in about 2 years...

I had an "old" Thinkpad laptop (380ED), it recognized my PCMCIA NIC (3c589D) but only after getting an "experimental" pcic driver from a Sun engineer... This was back in February 1998. Since then, 2.6-5/98, Sol 7, and Sol 7-3/99 were released, and the fix didn't make it into the distribution. From what I heard, apparently, 2.8 won't have it either !

I was extremely excited about Sun's initiative to provide the "free" Solaris for personal use. I though about the thousands of students around the world trying to learn a real operating system, how to administer it, and how to love to develop on it... I though about them getting jobs in a couple years from now, and pushing Solaris to be "THE" Unix of choice in the 21th century...

I though about the great momentum Linux has caused, and how lots of power-users get away from Microsoft Monopoly, and use almost a real OS. I though it could be great for Solaris as well, but sadly, hardware support is making Solaris x86 "the OS/2 of the 2000s". It's crucial for Solaris x86 to survive to have a device driver base like Linux has.

I'm a "Loud Sun Bigot", and wouldn't recommend anything but Sparcs as servers, however, I think there's a place for Sol x86 on the desktop, and as workgroup servers.

As I work a lot at home and on the road, I need a laptop, and due to lack of hardware support for high end laptops, PCMCIA NICs and modems ("The network is the computer", remember ?), I'm forced to use M$.

I think it's great to have the XInside guys support Accelerated-X on Solaris, I can live with having to pay around $200 for getting superb video support, but there's nothing like this for core drivers.

I don't think x86 is a threat to Sparc, it's just not there yet... and don't think Merced will make any difference. What CAN make a difference is to be able to use the exact same OS on both you server and your workstation/laptop.

During the last year, with the need for increased computing power, I tested 5 different Pentium-II laptops to check for Solaris compatibility... With no real success up to now. Next in the pipeline is the Tecra 8000... ALL THESE HAVE THE SAME LEVEL OF SUPPORT ON LINUX AS IN M$.

It would be great if Sun wholeheartedly supported Solaris x86, but I know it's a hard (and not very profitable) job, considering the overwhelming quantity of x86 hardware available out there... I understand Sun cannot develop drivers for the myriad of x86 hardware available, and the numbers presented to IHVs are not that exciting like for Windoze or Linux... But I think there is an alternative:

SUN COULD RELEASE CURRENT DEVICE DRIVERS' SOURCE CODE. I'm not talking about "sample" drivers' source, I'm talking about the real thing...

This way, we could get into hacking the drivers source for our "unsupported" hardware, and make it happen... As it was in the Linux community for the last 5 years...

WE CAN MAKE IT HAPPEN! WE NEED JUST A LITTLE HELP FROM SUN!

Solaris is so much a "real" OS when compared with Linux (specially when SMP is involved) that it'll be really bad to loose it. [Hmmm... OS/2 again ?]

Then there's the software... Q: Why isn't all Sparc software supported on x86 as well ? I mean, most (properly written) code, it's just a matter of recompilation (and a little increased cost in QA).

So I think the answer to the above is easy.

A: It's Sun's actions that, by not committing to x86, give a bad example to ISVs. Why isn't Java Workshop 3 delivered on X86 at the same time as Sparc ?

Come on guys, you have an incredible ability to deliver great technologies (both hardware and software), and it's much harder to believe that management/marketing doesn't seem to realize this, and do nothing with it.

I'm glad Star Division ported StarOffice to Solaris. But many other ISV should feel Sun's push and commitment to the platform. Since Sun appears to have "dropped" support for it (except for the lip service on x86 servers), these recompiles just don't happen.

Good things is that Sun still has people like Bob Palowoda, Casper H.S. Dik, and others around to help developers with Solaris. Remember that developers tend to "push" most the platform they like most. As long as it's Solaris, it doesn't matter if it's x86 or Sparc... It's just a rightsizing decision, which, mostly goes towards Sparc (or starts with an x86, and grows later to a Sparc).

In our company, there are TENS of developers, System and PreSales Engineers, all stuck with old laptops... Some of these are already looking at Linux, as they need more powerful laptops... Such a waste of time !

I really hope that Sun's TOP management has a chance to look at this, and finally decide to do something with Solaris x86, and put all their muscle like they did for Java, etc.

NOTE: All opinions expressed herein are my own and not those of my company, unless explicitly stated.

--

Pablo Ariel Kohan

Software InfraStructure(SIS) Division Manager

MEMCO Software Ltd.

I may have missed someone else posting this, there's a lot of response and I haven't read every one. I'd first like to say that I'm very grateful to see these posting fly around, Solaris is a great os and would be a welcome advent to the x86 world if only supported in a more dedicated fashion. Anyway, coming down off my soapbox, one thing that will be very difficult to find but well worth support would be software to take advantage of CDROM Writers.

My 2 cents worth. jb

I'm currently running Solaris 2.7 on my PC at home. The System configuration is a followed:

1. Motherboard - SuperMicro P6DGE Dual processor. Currently have only 1 processor running at 400 Mhz. USB on board.

2. 128 MB memory

3. Have S3 graphics Adapater

4. 17" ViewSonic 17GS monitor

5. Disk Drives are IDE Samsung 4.3 GB and JPS 2.0 GB

6. Netgear NIC FA310TX

At work I have a IBM Thinkpad 600 with 228 MB of RAM and 2 pcmcia 3Com adapter, which can only be used one at a time. One is a 3C589D and the other 3C574TX. I'm running Solaris 2.7.

On the desktop and my laptop, I can't use the USB port for any devices what so ever. Why, cause noone has built any drivers for the these. Specifically: QuickCam by Connectix which is the color version. My problem with this I see that Microsoft and Apple already widely using this BUS and devices for USB why isn't Sun. Taking a look at it Don't cut it.

As for the PCMCIA, I've only used it on my laptop. I'm tired of having to insert my ethernet PCMCIA card everytime at boot. This is a nice effort on the person part whomever wrote the pcic driver. Sun didn't even come close to getting it to work. Maybe why this isn't working because the unpopularity of the PCMCIA on sparc. I haven't even seen a sparc system that even comes with one. I've been using sparcs for a long time. Since 1988. That back when VME bus was still in big use.

Laptop support in which you are not supporting sucks. I dont' see any difference between a laptop and desktop system other than it is lighter to carry around. To me it is still Intel system which should be supported. I use it so that I can port my shell scripts to sparc and make very minor changes in order for it work. I also use the laptop to copy over programs which will run on the sparc but won't on the intel. I know that copying program from the sparc is doable instead of using Linux.

Don't get me wrong, Linux is a good OS but I like sticking to what I know best which is Solaris and not another OS even though I probably think that was a SunOS implmentation. Now I guest you might consider bring that OS back to Life just to compete with Linux. It is simply a BSD implementation like the old SUNOS used to be.

Also I don't see a reason why you guys can't do a port of some of the applications that Linux has done. Who knows you might get Solaris to run on a DEC Alpha.

Also I would like to see Sun Volume Manager run on x86. Sorry Guys but I don't like DiskSuite at all. I'm far too used to Veritas and Sun Volume Manager.

I personally will not use the Intel for a Server in the Business world but I'll definitely use it for Desktop and especially at home.

Thanks for taking time to read this whomever you are.

Much x86 hardware is crummy and not worth supporting, so I don't mind not having support for oddball stuff. Support for exotic video cards isn't important - they're basically targeted at juvenile games that I don't care about. Support for *necessary* hardware is a different matter, though. 1) SCSI cards - Adaptec's are admittedly crummy, but there are scads of them out there, and better effort to keep up with their unannounced chip changes would be very helpful, even if that just means documenting in a timely fashion which variants work and which don't, and realize that some machines (eg. Dells) have these chips on-board. I wouldn't buy a new SCSI card from anybody but Symbios and DPT - their products are clearly superior, so you really need to make sure that they work well. Similarly, 3com's ethernet cards are screwy and I've had no end of trouble with them. I've had cards that work under MS-DOS fail to work under SunOS5, even though they're on the HCL. Support for 3com ethernet needs to be better, especially since many machines have it onboard. I have Dell OptiPlexes whose onboard 3com won't work with SunOS5 - I had to stick in a DEC DE500 card, which worked great except that the kernel ignores attempts to force 100mb/s mode. Now that DE500's aren't made any more, the Intel cards appear to be the logical successors - their product line needs to be fully supported.

Sound cards - sound isn't mission-critical for me, but I would like to be able to view occasional animations with xanim and hear them. I wouldn't expect the various oddball sound cards to be supported, but the market leaders like *all* of the SoundBlasters really do need to work, and not just a couple of ancient ISA versions.

As others have noted, laptop support is critical, and on these machines, we don't have a choice as to which video hardware is used. We've got Dell Latitude LM's with beautiful bright screens, but a goofy NeoMagic chip. Running X in 640x400x16color mode just doesn't cut it. Xfree86 isn't worth the hassle. Sun's X server really needs to support the built-in graphics on popular laptops. I'm nothing short of amazed that with SunOS 5.7/Solaris 7 laptop video support has been dropped. Many of us can make a case for the superior quality of SPARC desktops (well, except for the embarassment of the Ultra 5), but few can justify paying a premium of $10,000 for a SPARC laptop, especially when the handful of SPARC laptop vendors are too lame to keep supporting new OS's on their hardware. I could make very good use of a laptop machine - we've got unused Toshiba Satellites and Dells - but they don't do me much good without a real X server, and it's tough to gamble on xig's. And I'd really rather not have to dink around with a Linux or *BSD. The result is that we end up having to run MS-DOS on laptops, which is really lame.

Perhaps Sun should pick an x86 hardware vendor and pledge to have Solaris x86 driver support for all of the cards / boards / etc. that the hardware vendor uses. It would be great if Dell and Sun collaborated on a "Solaris-ready" line of x86 hardware.

Get more better people to write there documentatiion not that "System admin guide by Winsor "

Get third party to write teach your self guides, and self study books for certification becuase there are alot of people who want to learn but cant affrd $2k for sun classes

You can get any book for microsoft products but you can't for sun Why Scott ?

Sun has to realise that individuals would rather run X86 than sparc for cost reasons

I've been developing applications for Solaris for the last six years on Sparc platforms. I've wanted to run an x86 version of Solaris at home for convenience. I've had, on a large number of occasions, customers ask to run my applications on x86 machines(in house already, more cost effective, etc). I've also had the desire to run demo's on x86 laptops.

But, none of this has been feasible given the lack luster support Sun has for the x86 systems. Even on my home systems I was unable to get Solaris running acceptablely. First it was lack of drivers for the SCSI cards I had. Then it was lack of drivers for Rendition chips, Riva chips, and NVidia chips(cards I currently own). Of course trying to use 10+ gig drives was a waste...And what about OpenGL support for x86?

In the end I was able to get a single test machine up using old/discontined hardware. The graphic performance is terrible as a result. And to add insult to injury, Solaris drivers for the ATI Mach64 card do not support DPMS!? To me, this isn't support. Yes, even XFree86 has DPMS support for this card.

I am all for Sun putting their driver support in to the Open Source movement. At least then there might be some hope of using hardware that was made less than 5 years ago. Developers could then submit their drivers to Sun for testing and possible integration into their OS.

I see the possibility for a strong demand for a stable, well respected, commercial Unix running on an x86 platform. If Sun doesn't step up to the challange something else surely will.

It would also be nice as a developer/user of the OS to be able to have a forum to post and ask questions that had representation by Sun. Usenet news is a great place for administration type questions, but I have found for specific coding/OS issues it wasn't as helpful because the people who knew the answers(i.e. Sun) weren't there to help.

I understand that IDE hard drives greater than 8 GB are not supported on the x86 version of Solaris. Yet, the same is no problem on x86 hardware (Ultra 5 and 10). Conclusion? Sun knows how to fix this problem but just hasn't bothered.

Others have mentioned lack of drivers for video hardware. This is definitely a problem. Also, I have a SUPPORTED video card (Matrox Millenium II, PCI version). However, the Hardware Compatibility List says, "Repeated switching between graphics mode and text mode may cause the system to hang. If this happens, reboot the system." First, this is the kind of crap I expect from Microsoft, but I expect better from Sun. Second, "repeatedly" can actually mean "twice in ten minutes". This can easily happen if I login and logout twice in ten minutes. (I logout when I leave my desk for a long period of time.) Third, Solaris is supposed to be a server O.S., and yet one of the most common video cards in the PC universe can cause it to hang. Fourth, this is actually a hardware bug, but a workaround exists on every other platform I know of (Linux and other XFree86 systems, Windows 95, 98, and NT) that supports the same card.

Again, after searching on the SunSolve web site, I have reason to believe that Sun has developed a fix for this but has not released it. It may be integrated in Solaris 8, but that is not helpful.

I would like to suggest that Sun would do more to make fixes available. Driver updates should be released about once every two months, if not once per month. Even if only one new driver is included or just one bug fix, it would increase confidence. Also, I believe Sun should have a driver status page on which new hardware support is announced before it is released. For example, if support were being added for a new video card, as soon as the development effort were far along enough to have a good estimate of the completion time, the effort would be publicly announced along with a release date that Sun is committed to. This would make it easier to make plans for hardware purchases. Right now, there may be a slew of new device drivers under development or none at all, but customers do not know what they can expect to be available.

Finally, this is not a hardware issue, but right now it is possible to serve Answerbooks off the CD-ROM, but it is very painful. For instance, the same scripts cannot be used to start and stop the server as one would normally use. It's no big deal for established Solaris shops (which can devote a disk on a server to serving Answerbook pages), but it's a big irritation to those who are trying out Solaris on a single test machine. (And you could say, "just get a big hard disk and don't worry about it", except that big hard IDE disks aren't supported!)

Having said all this, I use Solaris x86 as my primary operating system, and as a professional Unix administrator, I would be willing to recommend it for certain uses (such as a web server), especially if better hardware support were demonstrated.

Of course, as so many others I like to see bootable CDROM's, better PCMCIA/notebook support, AGP video driver support, etc. One thing Sun could do is open source/community source their x86 device drivers (not including the 3rd party proprietary drivers). Building a new driver fromm an existing/working driver is so much simpler. Besides, their device driver source doesn't really represent much market value given that Linux, *BSD have much better hardware support.

Most of what I wish to say has probably already been covered in great detail by others. I can only hope that Sun starts paying attention to the end users of their software and stop treating it the way that IBM treated OS/2. A far superior OS to Windows, but went no where because no one believed that IBM itself was really interested in supporting thier own platform.

Now that Sun may be aquiring Netscape's server product line, we can only hope that Sun will see the light and release versions of server software for the Intel version of Solaris.

As far as the lack of hardware drivers, I agree with all the other folks.

But we also need software, developers aren't going to jumping to Solaris on Intel unless they feel confident that Sun is going to support the OS, and therefore, its worth thier time and effort to do the same.

alan

I'll keep this concise. Most of the points have been mentioned already. Here's one simple wishlist:

* Basic video: sun MUST SUPPORT 256colors as a "generic". After all, CDE looks lousy in 16 colors

* high-end video: sun MUST SUPPORT high end graphics chipsets, and give good OpenGL integration for them. Otherwise, it cheapens the look. 90% of marketing, is looks.

* Monitor entries for Sun's own monitors?! I'm having a heck of a time getting a GDM17E10 to work in high-res. I have the appropriate cables, but finding the magic frequencies for 1152x900 is difficult. (VGA works)

* Bootable CDROM. yes, yes, yes, please

* BACKWARD COMPATIBLE. Sun should not even release a new version of x86, until it has already been certified to run on at least 50% of the hardware of the prior version. Solaris 7 didn't even come close to Solaris 2.6

* similar to above: dont get too clever with install, and REQUIRE plug-n-pray. 2.5.1 boots up great on one of my systems. 2.6 and sol7 LOCK UP on probing for devices. Let me have a manual way of specifying things for install.

* As for PPP support: Sun is welcome to improve and integrate my ppp script. Others are welcome to try it :-)

http://www.bolthole.com/solaris/configppp.sh

should work on both sparc and intel.

Would be nice to include possibilities to upgrade the boot diskette on the sun website. I don't want to get stuck with my older hardware, sometimes you just want a new piece of hardware. The NCR 875 is very well known and widely used and also supported in all linuxes I know. So why not support it?

Actually - this is more about software, than hardware. I have used Solaris X86 since 1994 and presently use it at home (thanks SUN for the free copy for developers). I like it, but it could be so much better!! And, there is always this nagging problem - is Solaris X86 for real - and it will be just dropped - like WABI! Hey SUN, how about treating them equally! For example:

JavaSoft - Do they know of Solaris X86? Try getting Hotjava, Java Web Server, latest JDK stuff - thank god I saved the Win95 partition! SUN has made Windows the preferred JAVA development platform!

Even Sunsoft - their Workshop compilers cost the same but lack memory checking features in X86

I used to use Solaris X86 on a laptop (an old Toshiba 75MHz Pentium) to demonstrate our SPARC solaris software. What are the chances a new laptop will even display a CDE screen w/o spending another $200 on a 3rd party X-server. And then, there are PCMCIA and LS120 drives and so many other stuff that won't work.

Dropping Driver updates - BAD idea.

My suggestion to SUN:

1. Get your on house in order - every single SUN software must be available on both SPARC and X86

2. I know there are a huge variety of hardware, so just follow a few well known brands, and follow them well. For example, how about announcing a Policy that from now on SUN will support within 3 months of release: a) All SCSI cards from Adaptec b) All Video cards from Diamond/Matrox c) All motherboards/CPUs from Intel d) All desktops from Micron e) All laptops from Toshiba/IBM f) All servers from Dell

Granted this choice of vendors are arbitrary, but as long as one set - any set - is available, we can field Solaris X86 systems with confidence. Right now, it is awfully hard to order a "Solaris compatible" PC from a mainstream vendor. (sure there is EIS and a few others)

SUN, once people see the stability (30day plus uptime) and performance (almost twice as fast as Windows NT as afile server - running samba), they will probably buy a sparc server. But to show it to them, you need to able to take one of their NT servers, and install Solaris on it!

Yes - Solaris X86 is a demo copy of the real thing - but not too many shows are happening!

Wake up SUN - see why Linux is displacing NT in the small business server arena. You could be a major player too! Not many small companies will jump straight into SPARC hardware without testing the waters (with a Solaris X86 PC) first.

Best of Luck!

Video cards. We need more drivers than the measly ones now available. Jeez, how about a default driver for 256 color 800x600? Linux and FreeBSD have this generic default so it can't be impossible.

I want to see the Solsris x86 CD in "El Torito" format, so I can boot with it, instead of the two stage floppy/CD boot, and then I would like to get x86 to boot over ethernet with no boot floppy. Neither of these requests is rocket science: Net boot formats and methods are well documented. I considered just writing the boot myself, but what happens when I upgrade? El Torito is also well documented: not a hard problem either. A good assignment for some hot-shot x86 type person as a first "real" job assignment with a little help to keep him/her from going overboard. By the way, net boot support was mostly in 2.5, but onlyl for SMC EtherEZ cards. I could live with that, but it was missing some small pieces.

I could sure use some support for x86 Solaris: To quote from an old Janis Joplin tune: "My friends all use Linux, I must make amends. Oh lord, won't you buy me a x86 Solaris.

Michael Lyle Inhouse developer of Motif apps that are the GUI for the machines that make the color shifting ink on US currency. (We use SPARCs for this task, but I would love an x86 laptop for debugging.)

I work at NASA on the International Space Station Project. Solaris has been chosen to run on some of the space station laptops. My project (there are other space station projects besides this one using laptops) has purchased something like 500 laptops so far, using Solaris. (Doesn't mean there will be 500 Solaris laptops on the station, but with testing, development, support, etc. what we use adds up.) We have real problems upgrading to the latest Solaris version becuase of device drivers. I am trying to get time to write drivers. I may be asking for more work than I can do, though. We need one specific device driver a Mil Standard 1553 PCMCIA card, my partner has been working on that. It is a real problem. We can only run Solaris 2.5 so far because of this driver alone. We need other PCMCIA drivers too, many of them sort of standard devices, like ethernet, flash cards and others. The drivers are keeping us at 2.5.

I know the amount of Solaris we buy is not impressive, but this is just one little story out here.

Dear Sir The struggle for market share is the struggle for MIND SHARE. Especially the mind share of the young techies who are learning about computers. When I did a contract at Sun in Toronto I asked about the price of Solaris on intel they told me about a THOUSAND BUCKS. Don't make me laugh. Recently I bought Solaris 7 for about 12 bucks plus shipping. Much better.

Why not make the server disk available? More programs means more stuff to learn about. For example Netscape is on the Server disk, but I don't get the server disk.

How about a CD of Solaris on Intel shareware. I can see that you might not want to give away your compilers, so put GCC on a CD. By the way I will never pay $1000 for a compiler. Try $100 or less. Finally, how about BRAIN DEAD EASY dial up connections!!!!!! I use NT as my gateway because it is EASIER TO SET UP. How about a proxy server / firewall that is easy to set up. Think about all the things that Linux/Red Hat has been doing right. It could have been Solaris.

Best wishes and GOOD LUCK

What we need where I work:

1. The ability to run Solaris on the existing laptops (IBM Thinkpad 386XDs) to demo our software to prospective clients. Solaris 7 x86 installs OK, but with no video support for the neomagic chipset used in the Thinkpad (and several other comparably priced laptops) Solaris is useless. I either have to buy (at $250/machine) a driver from XI Graphics (with no idea if it will be updated when the next release of Solaris comes along) or blow off CDE and use XFree86. Faced with that pair of choices, RedHat Linux 5.2 (which runs on the Thinkpad, has XFree86 already there and working becomes a better choice for us.

2. The ability to have Solaris 7 install and work on some of our existing Desktop systems (Intel Seattle motherboard, 128MB RAM, Pentium II/350, Matrox Millennium G200 Graphics, 10GB IDE HD, Ls-120 ATAPI floppy (Controller 0 slave), and Toshiba CD-ROM (Controller 1 master). The install dies in the midst of reading the floppy, with "cannot mount root filesystem". I've heard from others that this diagnostic means "your CD-ROM is not the slave device on controller 0". Now, I could add a regular 3.5 inch floppy, recable the CD-ROM, etc., but why bother? RedHat 5.2 installed without a hitch on this configuration. So we now have RH 5.2 on several desktops and laptops. That leaves the servers and two desktops running Solaris. Given that those are long in the tooth, and Intel platforms are much less expensive (and would work just as well in our environment), there is some pressure to get back to a single OS, and its not Solaris. The counter force is some long time Solaris users (like the CEO and COO) who prefer that OS, but even they are being tempted by the economics of the situation.

Bottom line for us is that Solaris 7 x86 had better clean up its installation act quickly (I could deal with replacing the G200s on the desktops with a "supported" video card, and maybe spending money on the laptops for the third party driver), or its going to be squeezed out of our company.

My 2 pence, Ric Anderson (working for, but not necessarily speaking for Oasis Research)

I work at a server farm where we run about 300 Linux boxes.

I have a Linux server and a Solaris x86 box here (I telecommute). As much as I like Linux, and as much power as it has, Sol x86 is my preferred platform, simply because of the ease of installation and maintenance--pop in the CD, install, configure, and you're up and running.

Right now, Linux is in a position to knock ms out of its position of dominance. Solaris x86 could be in that same position if it were promoted and marketed correctly--it's a great product, but the general public, those power users who are looking for an alternative to ms, don't know about Solaris x86. It's time Sun got out there and let people know that Linux is not the only choice they have.

Thanks a lot for making the Powers pay attention to what we x86'er's have to say!

I have been trying to load a promotional Solaris2.6 developement kit for X86 since last June. So far I have tried 3 cd drives and a new motherboard with no luck. I am now slowly buying everything right off the hardware certification test. The biggest problem so far is that all the hardware on the "hardware compatabilty list" has been discontinued. I looked into laptops and that was case there also. Today I will try a scsi card (was using IDE), and a scsi cd drive off of the certification test for the TX97-LE motherboard. If that doesn't work I will have to go to a scsi hard drive off the same list. I am not worried about my non-compliant graphics card (Mirage for fixed freq HP) as I can use Xfree86 server.

When I originally called Sun, asking for a new cd thinking the original was damaged, they refused saying the 90 day install support was over.

Well it has been almost a year now, and I am still at it (vacation, back injury, subsequent back surgery, and hardware compatabilty problems) and I am concerned that Sun will not send me the license file for the RTU license I was sent for the Visual C++ Workshop. This was a promotional commercial development kit sent to me from Sun and no one at Sun seems to know about it.

I have worked on Suns since 1989 and have a lot of respect for Sun Microsystems, which is why I want to run it on my PC, and use it for C++ and java development. I have been running bsdi unix since Beta, and I understand that the X86 platform is their mainstay while it is not so for Sun. I hope to finally have Sol for X86 installed and up this week.

Dear Mr. McNealy and the Solaris-x86 Management Review,

Sun is squandering a large market opportunity by inadequately supporting Solaris x86. Potential customers are forced by Sun's lack of both software and drivers into using NT, SCO, or Linux. Benchmarks prove that Solaris is by far the most efficient SMP operating system for x86 hardware, but few x86 customers study benchmarks. They typically just want a system that works. Unfortunately, in many cases, Sun has chosen to deliver a Solaris that won't.

For a company whose justly proud slogan is "The Network Is the Computer", the Solaris ppp package is a humiliating slap in the face. A fair-minded comparison between setting up ppp on NT and Solaris can only conclude that Sun dial-up technology is light-years behind. A pathetic fraction of the posts to Solaris newsgroups concern this longstanding and easily fixed situation.

Solaris x86 users tend to be software developers working at home or on the road. Leased lines are not an option, ppp has to work well, or another OS must be used. All but the most fanatic Solaris bigots are driven off by this one little problem.

Trying to purchase server-grade x86 hardware on which Solaris can be installed requires a customer to forgo many popular pieces of modern hardware. Solaris is well behind Microsoft in supporting new SCSI controllers and networking devices. Customers must carefully choose obsolete hardware to get Solaris to run at all. Since these controllers are often integrated with the motherboard, customers often can't use up-to-date server motherboards at all. Lack of support for available hardware also forces customers to select an alternative to Solaris.

Let me guess that Sun has invested many tens of millions of dollars developing SMP on Solaris so that it is 20 %- 30% more scalable than NT at the four-processor level typical of x86 systems. Lack of hardware support means that a customer can actually get better performance by just getting NT, because it at least supports the SCSI controller found on the 100MHz motherboards supporting the very latest x86 chips!

In the last few years, there have even been cases were the vintage hardware that Solaris does support is not being sold anymore!

Official Sun policy has eliminated laptop support. Technical travelers are forced to learn the operating systems of Sun's competitors. Once these travelers port their applications to Linux or SCO or NT, Sun's loss of revenue cascades exponentially beyond the single laptop sale that precipitated the port. The customer that the traveler is visiting sees the application first on a non-Solaris system. Sadly, this is happening even with Sun's own employees! Sun employees are forced to use Linux or Windows when travelling to a customer's site! Measuring this loss by counting the actual number of Solaris laptop sales lost is only the tip of the iceberg.

Closing the laptop gap might be as simple as striking an arrangement with a company that already has made the investment to support the latest video and CardBus technologies on Solaris. How much can that cost?

Or just make laptop device drivers open-source. Laptops are obviously in a different category than servers, so the distinction can be clearly made in the marketplace without damaging the server reputation that Solaris has earned. Even if the laptop device drivers were not up to Sun's reliability standards, there would probably be a net benefit for the reputation of Solaris. Sun can be open source where it doesn't put server reliability at risk!

The popularity of Linux proves the viability of the x86 marketplace. Is the price of driver support, a ppp upgrade, and a laptop deal so high that it justifies competitive default?

I would like to put a vote in for a GUI ppp. I found it too much of a pain to set up ppp under Solaris x86, so I was forced to use Linux as my NAT machine and just use Solaris as my workstation. I would much prefer to just use Solaris on all 3 of my boxes. I am using Solaris at home because of the stability and so that I may learn Java. I am forced to use ppp because I don't like cable modems and DSL is not available in St. Louis, Mo. just yet.

Thank you for you time.

The Solaris 7 (and Solaris 2.6) x86 can only see the first 8000MB on any IDE drive. Considering that there are now 20GB drives that are common, this is ridiculous for an OS that is called a "Server" OS.

I had to install another complete system to be able to use a 19GB IDE drive because of this lack of support. As soon as FreeBSD gets more reliable, I will be removing the Solaris system. This is amazing since I have been working on almost only Suns for 7 years now.

Sun's support in Solaris x86 is summed up with "Where is acrobat reader?" If Solaris x86 was important to Sun, then Sun should be working to get the very basic software required for use available.

Its a shame after buying my laptop and the listed supported PCMCIA card (3c589D) that I found the thing didn't work under 2.6 and then in turn 7. Spoke with Sun support who suggested various things such as force_8_bit mode et al... Eventually I decided to try a cardbus ethercard, so I went and bought one for a considerable sum of money. Again no joy!! It took some before Sun support said Oh yes, you need a patch for cardbus support, so I asked "Oh good - When will that be released..." to be told - "We've dropped support for laptops!!!" Eventually I forked over the 189 UK pounds for the XIRCOM parallel port ethernet, which granted does work, but I would sure like to get to use my 10/100MB PCMCIA cards!!

The crazy thing is that this is now the standard for PC's the old 16bit PCMCIA is dead everything is CARDBUS, yet Sun Don't Support it!!! They do the free Solaris offer, which is great but they've shot themselves in the foot by not having any CARDBUS support.. I suspect all those users that might have taken the free solaris offer up will simply switch back to linux or Windows and slag Sun of - Surely not what was intended by Sun's Marketing. Really what would be nice would be support for the realport cardbus card from xircom with the modem and ethernet.. Support that and I'd be very happy.. Ecstatic. I guess Sun like everyone has a resourcing issue, but what about co-operating with one of their partners like SNI, or Fuji who make loads of PC's and would be interested in support for CARDBUS. If the change weren't to extensive how about releasing the source to the existing PCMCIA drivers so the community can use those as basis for drivers for the other cards.. That would be Good and might help Sun out if there not interested in laptop support.

Last year I wrote John McFarlane, President of Solaris Software, asking him to reconsider Sun's decision to withdraw laptop support. I cited my experience trying to find a laptop to demonstrate software to a potential customer software. This software currently runs in production on a Sun SparcServer 1000. While Mr. McFarlane seemed to sympathies with my frustration in trying to run Solaris for Intel on a recent Toshiba laptop, he explained that driver support for laptops is cost prohibitive.

I understand that driver support is expensive and that laptops present a particular challenge because of the varying types of bus controllers and the myriad of PCMCIA and card bus devices. However, Sun must maintain at least some support for these "low-end" systems. Intel-based laptops, once only used by managers and sales people, are now part of the tool-set of the technical staff who administer and develop applications for Sun's Sparc server systems. Does it not make sense to allow these Sun customers to run the same powerful operating system on all their systems?

Sun has made it clear that it would like to focus on the Intel-based server market, particularly the servers from its Solaris System Partners. I was pleased to hear that it intends to also support the current workstation and server offerings from the high volume Intel vendors like IBM, Compaq, and Dell. Recently I was asked to set up a Dell PowerEdge 3200 workgroup server, the customer was a "Dell shop". This machine was originally to run Microsoft Windows NT, but I convinced them that the Oracle application would run faster and be easier to maintain under Solaris. While the machine was listed on the Hardware Compatibility List, last year Dell upgraded its disk controllers to the Adaptec 7890 U2W. To my surprise, I found that Sun did not have a driver for this controller. While I understand that Sun cannot certify every future configuration of Dell's machines, the Adaptec has become the most widely-used Ultra2 controller and is included with most dual Pentium2 mainboards. I know that Sun engineers are working hard to release this driver, but I must ask why as of the March maintenance update it is still not available? Perhaps because Sun has removed the Driver Request Page from the Solaris Support Web Page, it no longer can judge the demand from Sun's customers for device drivers?

If Sun is truly committed to being the enterprise computing company, it must support the entire enterprise: laptops, workstations, and servers. If Sun is truly committed to maintaining and growing its market share, it's management must listen to its customers.

As a professional developer and long-time user of SunOS and Solaris on both Intel and Sparc, I find that Solaris is the best development platform for my development. I also support HP and IBM, but I develop on Solaris first, then port to the other platforms.

However, it's frustrating to use Solaris on Intel, as there are so many things that it lags behind in, compared to Solaris on Sparc. We have decent hardware for the Intel platforms, but the drivers are not available for the major hardware vendor's products. For instance, we have standardized on Ultra2Wide SCSI disks, but although we have been running them for more than a year on Windows- based systems, there is no driver support for the Adaptec cards we use (2490U2W). This is an industry standard card, and without question should be on the HCL, but although I hear that support is 'coming' for it, there has been no driver in over a year. The request for support for this card have been long and loud on the list, yet it's still not available.

We also could not run Matrox's Mill-II AGP video cards, as the AGP video has not been supported. Again, these cards are industry standards, and should be on the HCL, but the support is just not there. It seems as if we constantly run one to two years behind the rest of the industry in hardware support.

Additionally, since Sun is dropping Adobe's Display Postscript, why are we still locked into Postscript- only printers? Sure, HP is a competitor, but they also make one of the most popular and reliable printer lines in the world. So where is support for HP's PCL? Most folks have to either buy the Postscript option boards for the printers (gee, doesn't that put more money in HP's pocket??) or use a shareware driver such as the problematic Ghostscript to handle the rasterization. It seems so stupid, as again, it's Sun ignoring an industry standard. Who the hell cares if they are a competitor? Getting the job done is what's important to the customer, not inter-company rivalry. Color printing? What's that? It seems Sun is not even aware that folks use their platform for Web development. I guess they can't imagine why someone would want to print to one of those 'dinky' color ink-jet printers. Proofing your work must never have occurred to them. Yet they are probably one of the largest web platforms in the market! So for now, we need to buy $10K color lasers with Postscript to get color printing. I've had to try to justify these for the yearly budget, and the boss just laughs, and says 'Use a Windows system for printing.' It actually is cheaper to buy a complete PC with a color printer, and network it to the Sun.

It's also been frustrating that for years we had the same old compilers on Intel. Sun rather frequently updates the Sparc compilers to handle the newer chips. However, the Intel compilers were not updated for about 3 years. Only recently was the version 5 compilers released, and they still are missing features that are standard in the Sparc compilers (such as memory bounds checking). I don't know what happened to the 'parity' we were promised with the Sparc platform, but it has always been a half-hearted effort at Sun.

We have been told again and again that Sun is serious about supporting Intel under Solaris, and promised parity with the Sparc platform. Well, no-one believes that anymore, as the tools and support are just not there. I would love to see a market develop for software applications on Solaris Intel, but since we can't get support for the current hardware, and can't get good software tools that match point for point with the Sparc versions, it doesn't make much sense to develop for Intel.

There is a chicken & egg problem here, and only Sun has the ability to start the ball rolling. I suspect that internal politics are what's holding things up here. Perhaps Sparc's product managers are trying to dynamite the chance of any competition from the Intel side, and are dragging their feet in any way that will impede progress. But in reality it doesn't really matter. What matters is what actually happens, and so far, that's not much.

Every one says 'Wait until the Merced chips come out, then Sun will really work on the Intel products.' Oh, yeah? We've heard that before too. It appears to be a good bet that Sun will only meet their contractual obligations with vendors like NCR, and just has no interest in getting Solaris on Intel moving into the 2000's.

I do commend Sun for their recent promotions of Solaris on Intel, providing low-cost access to the OS for personal use. This may start a groundswell of interest in a stable, robust OS, and let people know that there is an alternative to the Windows platform. But until there are thousands of application programs for the platform, it will not see the widespread use that it could achieve. And until the support and feature set of the tools offer a decent development platform, the applications just won't appear.

There is so much that Sun could do to help this. Open sourceing of the drivers will help. But the biggest thing Sun could do is encourage or require that all Sun-approved applications be offered on both platforms. In most cases, at least for well-written software, this entails a simple recompile of the source code on the Intel side. That's not much to ask from a vendor, and I think most of them would be glad to do it if Sun provided some incentive. Surely Sun's marketing folks could develop a program to do this, and get the major software vendors to join in. Hell, look at Oracle. They have been supporting Solaris on Intel for years! And many other vendors do also, so it should not be a big problem to get a vendor program going.

A few other gripes:

Sun should stop messing around with the bundleing of their OS packages. I have many support contracts with Sun for software, but every time we go to upgrade an existing package, they have changed what the package contains, which requires us to purchase additional software. Hell, that's why we have a support agreement. Where did the Internet extensions go on the current server package? Now it seems to be a separate product. Sun also did this a while back with the compilers. We had a large number of compilers under support, so that we did not have to repurchase them each time an upgrade came out. So Sun decides that the newest package is different enough that the support contract does not cover the upgrade. Huh? Who's bright idea was that? Do they intentionally do this to irritate their loyal customers?

And where is the advertising for Solaris on Intel? We used to (some years ago) see serious ad campaigns for the product, but I haven't seen any thing for a few years now. We turn on the TV, and all we see are ads from IBM for their net servers, or Dell and Gateway hammering us about Windows PCs. Lets see some ads targeted at the ISP and business world. Of course, we need some apps to run. The chicken and egg again...

Still, the recent purchase of Netscape with AOL should see some decent enterprise class applications coming out on Intel. Or is this another parity problem?

And what's the problem with dial-up networking? Every other Intel-based OS that I can think of does this very well. But if you look on the list, probably 30% of all posts are request for help in getting dial-up networking to function. Want an exercise in futility? Try to get a Intel-based box talking to an ISP who uses NT servers for dial-in. There are web sites that do nothing but provide "how-to's" on this, as it's black magic. This seems ridiculous for a product made by a company who claims that 'the network is the computer.' Solaris really falls down here. Most folks rip out the ASPPP package, and put in a shareware substitute, because at least that works. I'd bet a bunch of bucks that Sun's product managers hide this fact from upper management.

I guess it all sounds like gripes here. And I'm sorry, but that's how I and many others feel, as when I think about Solaris on Intel, I just get irritated. It's so close to being a really great platform, but the pieces that are missing are crucial, and no one seems to care at Sun. It encourages developers to say 'The hell with it, lets just support Sparc', and it's hard to believe that that is not the intent at Sun.

Yet, all in all, Solaris on Intel does work, and work well. It has so much to offer to someone who needs support for multiple CPUs on Intel, and is probably the only OS that does this well. It's just that there are so many other problems that folks tend to ignore the great features of the OS.

There are hundreds of thousands of X86 users now. It's a market that is waiting to take off (and waiting, and waiting..) Sun should decide where it's going with Solaris on Intel, and either give the market up to Linux (which seems to be happening by default), or get behind the product with some real muscle, and make it a viable platform. It needs some programs like the "*old*" catalyst program, where Sun went out of their way to help folks get going. A highly visible 'Sun Certified' program for applications would also help, with some support for a packaging label on products that have been certified on Intel/Sparc. Some pressure from Sun to existing vendors to support the X86 platform could turn the few decent application programs available into the thousands that are enjoyed by the Sparc community.

Will this hurt Sun's hardware sales? I think the impact would be minimal, as many of the applications out there require enterprise-level hardware. When that's not the case, it would be far better to offer the X86 solution than to let folks move away to the Windows environment. At least you would keep the user as a Solaris user. As his/her needs develop, they would have all the reason in the world to move to a Sparc/Enterprise platform, as they are already familiar with the environment. You might lose some minor bottom-end hardware sales, but think of all the cases where you could take a platform away from M$-Windows.

I'd be very pleased to think this message might be passed along to Sun's management. It's probably quite characteristic of the situation from a developer's point of view. And perhaps upper management is just not aware of the situation. I think the best thing in the world would be for someone in management to be required to subscribe to the Solaris-on-Intel lists, and provide a monthly report to corporate management on the 'state of the union' of the platform. Based on the difference between what Sun's press releases say, and what actually is happening in the world, it would be an eye-opener!

X86 users are tired of being the 'poor relation' in the Solaris world, and would really like to see a change for the better. We all realize that it needs Sun's wholehearted cooperation to do this, and hope that they will see the light.

Sincerely, Carl Erhorn, Sr. Engineer Hyperion Solutions, Inc.

Note:

These are my personal opinions, and are not to be taken as an official statement on behalf of my company.

We are building two database servers under Solaris x86. We selected Solaris because of its stability and the amount of Internet-based support there is for it. Key point: we chose Solaris, not Sparc.

When I went shopping for RAID SCSI cards I was very,very disappointed with Sun because I heard the same thing several times from different vendors.

The ability to monitor and operate the RAID cards was available under most other OSs yet Sun had failed to port the support tools even when given the source! This is ridiculous. What good is RAID if the host machine has no idea that it is failing. It made me think that Sun maybe didn't want to see an x86 box succeed as a reliable server box.

Remember, not everyone has the money for for a nice multi-cpu Sun server. Those people are going to use NT. Is this what Sun says: Either buy our boxes or go ahead and use NT?

Other than the push for the support of higher end laptops, and as important the softwindows implementation that never 'happened" that bypassed emulation Windoze95 or "Nice Try (NT)" ... I'll let the rest of these good people (especially Bob Palowoda) suggest the needed hardware. I would also suggest that there is an un-tapped opportunity in the CAD niche with a toe-to-toe port of the current autocad and turbocad that Gates isn't smart enough to protect and as well, a current up to date version of ORBIX under X-86 Solaris.

As for us, I'm a D.o.D. Prime Contractor, and have been a Sun beta (and a vocal Sun bigot) over the years in the intelligence community. Your own people used to tell my that your personal pitch was, "well, if I can't sell you a SPARC right now... let me show you this wonderful operating system". We've tried to use that in the intelligence community for a long time (try your own Jim.Hoffman@east... , I'm quite sure he'll back me up as to my long term, and long time efforts on Sun's behalf). There are simply too many places to list where Sun could start eroding Billy-boy's corporate market share in our community if there was more tangible support for the Intel platform, as much at the desktop as at the server level. Particularly in laptop... to be able to haul your code on a known good platform... in your briefcase to have it reviewed, as opposed to hoping that there will be a machine in the bosses office... that works, is invaluable for the legions of programmer Govies, and contractors in our community.

HP doesn't know how to market Unix on their laptops... they've rolled over to (shudder) NT... Time for Sun, the innovator, the engineering company... to once again strut their stuff and help us save our selves from a world where there no choice but Gates.

I personally think that to do that you need to move the product back off the freeware shelf... and back into your revenue generating inventory as well and charge us for it... but that's my opinion. Dispensations for developers are a good idea (you're no stranger to that) and whatever other promotional ideas you care to implement. I realize that this will not exactly engender me to a lot of people on this list... but if Sun doesn't make money... they will not be around to provide us an alternative to Gates's garbage in the first place... so we need you to have a profit motive for enthusiastic support for X-86 Solaris. jccampb

We need the latest PCMCIA modems supported.

We develop software to diagnose problems with Gas Turbines. Our product was developed on Solaris and runs on either SPARC or x86. Whilst we have recommended SPARC hardware in the past, our customers will only accept x86.

Some engineers are mobile and need to monitor the turbines remotely, so they need high speed modems on their portables.

Please support Notebooks, especially the latest models.

Even if it just for the major brands such as Toshiba or Dell.

We develop software to diagnose problems with Gas Turbines. Our product was developed on Solaris and runs on either SPARC or x86. Whilst we have recommended SPARC hardware in the past, our customers will only accept x86. We need portables to demonstrate the product running on x86.

I would greatly appreciate if SUN could update the pcic driver or (as a last measure) release its source.

I have been working on SPARC systems for some time now, and I won't use anything else as a server. I think Sun should make the huge Intel/x86 market work for them. Keep adding drivers for common components in x86 systems. Laptops need more support/drivers if it's going to become a more common Solaris workstation. The company I work for issues Notebooks to all employees for remote access and as a main workstation in the office. I turned it down because they won't run Solaris. (I refuse to work on M$ systems.) I don't think x86 will ever survive in a service provider (it's not a server, it's a personal computer). Get more PCMCIA cards on the HCL. The 3Com Etherlink III is currently the only network card (PCMCIA) supported on Solaris. I'm not a fan of Linux, but even Linux supports 100Base-TX PCMCIA network cards.

I think the overall installtion needs inprovement. Anyone who's worked with IRIX will tell you about the installation process. While it's lacking as an OS in some areas, the install is clean and easy.

Make the compatability between SPARC and X86 work better. Now that PCI is implemented on SPARC platform machines, make drivers for the PCI hardware on the market. This includes the NIC's, Video Cards, SCSI Adapters and whatever else folks come up with.

Is Solaris x86 a desktop OS ? If so, how come it's so hard to find anybody shipping an Intel box with a video card that's (properly) supported by Sun ? What would be even better (from our perspective) is OpenGL support for 3D video hardware.

Also, network cards. If Solaris x86 is a server OS, then why is it so hard to find a 10/100 NIC that is supported and still exists ? If you're lucky you'll find a dealer that has some dusty old boxes on the back of a shelf somewhere... I know NCR and (eventually) Siemens have Solaris x86 servers but what if we want smaller development machines ?

PCMCIA has been covered by others. Solaris x86 laptops are useful for so many reasons - developers, contractors working on customer sites, product demonstrations.

Power management support is another obivous ommission. Again, mostly for laptops where you would want to slow the CPU, shut down pcmcia cards, screen and disks, but it's very useful for desktop machines too. Even simply being able to power off with ``init 5'' would be an improvement.

My comments below lost their formatting and are hard to read. If you are intersted in seeing the original, go to SolarisX86.txt.

more than one pcmcia ethernect card needs to be supported 100mbs pcmcia support is also needed not to mention the overall lack of concern for solaris x86 on laptops in general - i don't care if the number is limited. pick some and the stick to it.

every seminare i attend (the most recent was in dublin, ireland (i am a consultant working out of my own business in san ramon california, was full of sun engineers with toshiba laptops running solaris 7 which they were using for presentations - ironic since sun's position is that laptops are not supported in solaris 7.

Actions speak louder than words.

My background. I am a former part owner of a Sun VAR that made one of the largest early sales of Solaris x86. I solved a few of my own installation and hardware problems when I was installing Solaris 2.1 x86 that Sun's customer support couldn't solve due to their lack of experience with PC hardware. Once I had solved my problems and reported the solutions back to Sun support, I started getting phone calls, which I didn't mind taking, from them asking how to solve problems with Solaris x86 installations.

Several times, I had rather highly placed people from SunSoft in my office asking what it would take to get me to sell more Solaris x86. At that time, the big selling point, still one today, was if it compiles on Solaris SPARC, recompile it and it will run on Solaris x86. In these early stages, there wasn't the dearth of software available for Solaris x86, as there is today. There was no software. My answer to SunSoft was to put their money where their mouth was. Build Sun NetManager, Online Disk Suite and their other tools. Prove that you can actually recompile SPARC code and have it run on intel. Create software, and along with it a market, for Solaris x86 and then I might be able to sell more. But as it sat then, with the stability problems and separate code bases Sun had for the SPARC and intel versions of Solaris, I wasn't going to be able to ethically push much more of the product.

Since leaving the VAR business and going back to consulting, I've been doing a lot of system administration work, programming and teaching, all of which has either been on Solaris or in Java, currently I'm doing Java on Solaris. I couldn't even begin to tell you how much Sun hardware and software I've directly or indirectly had a hand in getting to the end user. At a bare minumum, it must be into the 10's of millions of dollars. Also, I evangelized many people about Java and am presently moving an old DOS software product of mine to Java to put a GUI on it, rather than port it to the Win32 API. So, even though I've had my rough times with Sun, I base my recommendations to my customers on what I ethically think is best for them. Many times this has meant recommending to them the purchase of Sun products.

I sit here at home writing this message on my Solaris 2.6 166 pentium, wishing I got move to my AMD K6-2 300, but alas, I don't have drivers for the Matrox g200 video card. So I've been thinking about puting linux on that box. We'll see, Xfree86 is sort of working under Solaris on my 166, so maybe I can kludge together an install on the amd, but I really shouldn't have to do this.

During my work, I have brought many people over to Solaris, mostly SPARC, and today I continue to do so. As far as the "big" iron goes, I can't find anything better to recommend to my customers than to go with SPARC and Solaris. I'm not quite so fast to push Solaris on intel. I'm just not sure, especially with recent events, as to whether or not Sun is going to continue to support it at all, let alone support it at the pace it needs to catch up with linux, Solaris SPARC and other operating systems and hardware. So, I use it at home, mostly happily, but can't get many other people to make the move. To be fair, much of my work has to do with larger systems, which PC hardware just can't handle. But a lot of desktops have been lost to Windoze that used to be Solaris on SPARC. They could have just as soon gone to Solaris x86 if my customers had confidence in the product.

Early in Solaris x86's life, the confidence factor was a big deal. The OS was young and a little rough and Sun made the mistake of splitting the source code base between SPARC and intel (just like those 386i boxes they used to sell). Several years ago, I was still pushing Solaris x86. It had earned its stripes and was a nice little OS, which it still is today, and Sun appeared to be very committed to it. Yes there were holes in the devices supported by it, but it was a great alternative to expensive SPARC hardware and linux was not as stable and well known then as it is today. Today, it is I who lack the confidence in Solaris x86, or rather in Sun's lack of committment to it. At this point, in many situations, I cannot ethically make the recommendation to my customers to install Solaris x86 or go to the expense of making their software available on it.

Frankly, I've gotten to the point in the last few months that I've had to admit that the only reason I haven't moved to linux yet, is because Sun is providing a good JVM for Solaris intel. The linux JVMs don't yet cut it for all the Java work I'm doing these days. But, then Java2 on linux will have a JVM, although it won't probably be as optimized as the Solaris intel JVM. On my faster computer, it should still run circles around the better Solaris JVM on my slow computer. This situation makes it much harder for me to justify all of the hassles and that feeling that Sun would just like me to go away that I tolerate while I continue to run Solaris x86.

Then there's Java3D. My goal in life is to eventually write games. Since I'm not much of a fan of windoze, Java and Java3D is one way I might be able to do this eventually for a wide audience. But then there's no Java3D for Solaris intel, even though I spoke to the Java3D product manager at JavaOne last year and pointed them to Mesa as a way to get OpenGL support on Solaris x86 that didn't require SMCC to give SunSoft something. It doesn't look like this is ever going to happen at this point. So fine, I'll live without Java3D, or buy that SPARC I almost bought last December, but then I shouldn't have to do that.

Ok, lets go get Java Workshop 3.0. After buying 1.0 and support, then upgrading to 2.0, I was anxiously waiting for 3.0 to fix many of the problems that made Java Workshop nearly useless. Finally, two months late, its released ... but only for Solaris SPARC with the Windoze intel version likely coming out in April and no mention at all that there will ever even be a Solaris x86 version. And yes this is largely a Java program that has a few pieces of natively compiled code, like a faster Java compiler. So why can't this be made available for Solaris x86 soon ?

Ok, lets get hotjava, the java browser. If you go to download it, you'll notice there is a Solaris SPARC version, a windoze version and a generic version. Should I be detecting a pattern here ? Is Sun trying to tell me with their Java software, of all things, that I shouldn't be running Solaris x86 ? I'm not sure what to think here and this isn't the end of it, I could find more examples, but I think this should be sufficient to frustrate any Solaris x86 user who is trying to do Java from Sun on Sun's operating system, albeit apparently the bastard child that Sun wishes would go away.

Then the other day, I got slapped up the side of the head again. Sun released a bunch of the source code, beyond Java2, to blackdown, including Java3D. Linux is looking alot better on that AMD box these days and I've got a copy of caldera sitting there waiting for me to find time to start the installation.

Sorry, I still can't believe the originators of Java can't make their Java programs, just look at the Java Developer's Connection in the early access session for at least 2 more than I've mentioned, run on their operating system, or worse won't take the time to build it to run on Solaris x86. These are some of the actions that Sun is taking now that are speaking VERY LOUDLY to me right now.

Ok, so I've beaten "write once, run anywhere" Java over the head with a stick and anyone reading this is probably sick of it. If you go to http://www.sun.com/solaris/downloads.html you'll find a few more and if you take the time to look through Sun's software catalog, you'll find even more. Some of these don't make sense on Solaris x86, but many of them do and each one that isn't available for Solaris x86 tells me how unimportant Solaris x86 is to Sun and how much they are, or aren't, committed to the platform.

Then there's the third party support. I'm glad Sun and Star Division saw fit to make StarOffice available on Solaris x86. But many other software publishers need to feel Sun's push to commit to the platform. Since Sun appears to be rather silent on this, and has a bad track record themselves, these recompiles, which most "ports" would be from Solaris SPARC, just don't happen.

So now, I'm left in a state of confusion. I really love Solaris x86 and have tolerated its lack of device driver support (I haven't even mentioned things like 3dfx voodoo cards which I can get support for under linux to name one of probably more than 100 more devices I'd have support for if I moved to linux) and can even understand the costs associated with the development and maintainance of the large base of device drivers needed to support a PC based operating systems these days. But I keep getting pinged in the world of "write once, run anywhere" Java because Sun doesn't support Solaris x86 here. Some of this lack of support comes from Sun's other lack of support, i.e. in the case of Java3D not being available on Solaris x86 because Sun's OpenGL isn't available on Solaris x86. However most of it isn't because of this. It is simply Sun telling me that they either don't care about their Solaris x86 users or that the installed base simply isn't big enough for them to support. This is the only conclusion I can draw from the situation and it speaks clearly to me that I can't expect Sun to continue to offer products for Solaris x86, therefore I can't ethically recommend it to my customers for any type of significant use.

I'll continue to use Solaris x86 at home for awhile, at least until I get linux installed on my AMD box and see if I like it. I won't be making any more recommendations to my system administration consulting customers to install it on their desktops because of lack of current device driver suppport and an unclear future for any new device drivers. Nor will I be making recommendations to any more for my customers to install it on their servers because of those same reasons and the fact that linux's stability is as good or better than Solaris x86 these days and its much easier to move someone from NT to linux, mostly because of the press linux gets these days, than to have to make a case I can't really support for Solaris x86. I also won't be pushing any more on my software consulting customers to make their products available on Solaris x86, which I've sucessfully done at times in the past. My customer comes first and I'm not going to commit my reputation to Solaris x86 any more than Sun commits their resources to their product.

And all of you guys who actually worked on Solaris x86, you did a great job. The operating system is very reliable and except for a few differences caused by the PC hardware, you can't tell whether you're on a SPARC or an intel CPU when you are running the OS. This is entirely a failure at the high levels of Sun's management to not commit to the product and their abandonment of it. With a little marketing and committment to support this operating system by taking positive actions, Solaris x86 could be making great inroads into many places. Under the current conditions, it will quickly and decisively be replaced where it exists by linux, NT and Solaris SPARC and where it isn't it just will never be.

Actions speak louder than words, and Sun your actions of late have told me I should move away from Solaris x86 because you yourself don't want to be there.

None of the modern graphics chipsets seem to be supported on Solaris x86.

I'm looking to upgrade my graphics card to better run Solaris, and I'm having difficulty finding shops that still stock anything on the compatible list. I appreciate that there are so many hardware manufacturers producing so many new products that Sun cannot possibly keep up. But I feel that support for the big selling major vendors' chipsets, like the Voodoo Banshee, is imperative - even if only in 2D. Many people, like myself, want to run applications like The GIMP, and generally have a nice responsive windowing environment. (And can't you do something about the lumpy cursor movement in X?!)

I think that Sun has a marvellous product in Solaris x86, and should exploit it to the full. Anything to undermine NT's dominance of the desktop would be great, and to do this Solaris must be capable of running on pretty much any new PC. Don't limit it to 2 years out of date hardware, or Linux will bury it.

As mentioned before, my biggest problem seems to be that not many people know about it. Many applicaations that are available on SPARC are not compiled on Intel. And it literally should just be a quick recompile. Even Sun is guilty of this. There are many applications on the Sun homepage that are not available on x86.

Many vendors think I am crazy or stupid and say things like, "Well, Solaris runs on Sun machines. You must be thinking of Linux or Windows NT".

I would love to recommend Solaris x86 a lot more than I do, but I am always a bit worried in the back of my head wondering if I will get "screwed" later on because something I REALLY need is only available for SPARC.

Please push the word about Solaris x86. Make sure all Sun developers know about it and how source compatible it is.

I've been using Solaris since a long time (before it was named Solaris, before turboSparc was born) and there is no doubt it is a great OS. No need to say it stands far away from NT. I greatly appreciated the port to x86 because this opens new markets to my company : there are many situations where we could not push Solaris as a viable solution because Sun HW costs too much for our customers. In those cases, customers choose NT Server + NT Workstation on Intel HW : it looks (I said "it looks", not "it is") easier for system administrators, it can run on high end Intel HW (servers) and middle to low cost PCs (desktop workstation with good graphics card and SCSI devices), it has 99% of the applications people need to run their business and those applications have reasonable prices. For decision makers, this is enough to choose NT over Solaris, from laptops to servers. I hoped Sun could achieve comparable results, that is allowing customers to run their business, from servers (maybe with Solaris Sparc for those) to low cost desktop workstations and laptop computers : in many cases, that would be a solution that proves to be cost-effective (Intel HW is cheap) and homogeneous (running basically the same apps on all hw). But I can’t sell that kind of architectures to my customers because Sun is not commited to support Intel HW (and what about Merced PCs ?) : nobody can reasonnably buy 2 years old PCs just for the pleasure to run Solaris. So what ? If Sun wants their customers to buy Sun HW + Solaris for high-end servers and let them buy NT for desktops and laptops, this must be said clearly and it will make things clear for everybody. If Sun wants their customers to stay in the Solaris arena, then this has to be done : supporting brand HW and encouraging others to port vital applications must be started (I can’t even say continued...).

In my opinion, this is a wining strategy : it will not grab a lot of market shares from Sun HW and it will certainly bring new customers to Sun HW after they experienced Solaris x86 and need more horse power than Intel can give.

So, why don't you make a move ? Go, fight, promote, encourage others, create new market shares, and HAVE A DECENT SUPPORT FOR INTEL HW : THE RETURNS WILL COME.

In general, I am quite happy running Solaris 7 on my x86 hardware. If it had more applications and supported more hardware, it would be a definite Win/NT killer. The things I am most happy about are:

* reliability

* robustness

* security (with ipf based firewall)

* multi-cpu performance

* configurability

* all the GNU stuff works as its supposed to

(luckily, most of the apps I need are here)

* I can't wait for Intel to supply 64-bit hardware

(I don't have the budget for Sparc, but I've seen it work)

* even the beta version was quite stable

Compared to Win/NT, it should be a no brainer run Solaris based servers on either Sparc or x86 hardware (depending on available budgets). The fact that so many people run NT on x86 hardware is testament to the fact that if you own the client, your server solution does not need to be that good. Conversely, without an adequate client presense, it will be difficult to sell in the server market, especially the mid to low end server market. In addition, more and more low end servers are being built using desktop hardware and the performance of low end servers already exceeds that of the high end servers of only a few years ago. To this end, low and mid range servers are desktops. From a technical point of view, the operating system of a Solaris based desktop is far superior to the operating system of a Win/NT based desktop, however, it is unlikely that the market will accept a Solaris desktop solution unless several problems are solved.

* drivers, drivers, drivers

There should be no need for the HCL except to say what works best. Everything should just work. If this means you need driver-drivers to be able to utilize native Win/NT or linux drivers, it should be done. If you need to bribe/bully graphics card vendors to supply X windows drivers, it should be done. If you need to find dozens of subcontractors to write drivers, it should be done. If you need to release all of the source for the drivers and let the user community contribute, it should be done. Of course, native Solaris drivers should be developed where it makes sense from a performance point of view (i.e. disk controllers, network interfaces, etc.), but when they are late (which is almost always the case), it won't matter since a degraded solution will exist. Any solution is almost always better than no solution.

* laptops, cardbus, PCMCIA

Laptops are the clients of the future. Need I say more?

* bios interactions, power management etc.

It would be nice if Solaris could interact with the BIOS, motherboards and other hardware so that advanced power management and other configuration and monitoring would work. A 'motherboard driver-driver' is what's needed here. There are absolutely no performance issues, so it can be as inefficient as possible (i.e. use Win/NT drivers).

* ppp

The traditional Sparc client plugs into an ethernet port and setting this up is quite easy, but most x86 based clients talk to a modem. A GUI based interface to set up ppp and connect to an ISP must be part of Solaris. This would dramaticly reduce the effort and expertise required in order to set up a Solaris based machine.

* applications

Why hasn't Sun jumped on the Wine bandwagon? Putting a dozen or so top programmers on this effort could turn out to be the best investment Sun could ever make. Software developers will not target applications to an X-windows API until there are a lot of X-windows capable desktops. Unfortunately, the most pervasive API is WIN32. Wine offers the possibility to bootstrap this process by mapping the WIN32 API into an X-windows API. The right way to think about the WIN32 API is that it is just another X-windows toolkit.

* software subscription

I am somewhat disappointed that I paid a good chunk of money for a 2-year subscription a few months before the price virtually went to zero. What's more, I don't seem to be getting anything above and beyond what you get when you purchase the software for just shipping and handling. Maybe Sun should toss a bone to Solaris x86 subscription customers?

Yes, I know this is an x86 list, but bare with me! I've been a Sun/Solaris bigot for years now, I can't think of a better hardware/software combo.

However, I think that it's vital that Sun keeps it's support of x86 hardware (especially laptops) up to date. The sysadmin where I work is a Linux nut, but I think we should be using SOlaris. His response (apart from the usual source code availability - he's not even a programmer!) is lack of driver support, so I think up to date driver/hardware support is crucial.

Personally, I prefer to use SPARC based machines, but I understand that in a lot of cases, they can't be justified on a cost basis. In these circumstances I think it's better to have SOlaris x86 on the machine than some other Unix (or even, God forbid, Micro$oft products, which incidentally, I refuse to use).

I hope this campagn has the desired results!

Would like to see Sun getting behind other IHVs and partnering up to provide Solaris x86 as an alternative (to NT, or even Linux) on Intel based servers?

No question that Solaris x86 has a big advantage against the competition when it comes to SMP, and Sun should use that to their own advantage and start reselling Intel hardware or making a stronger presence with other vendors who do provide the Intel hardware for sale. Sun may find that it can be as profitable to resell Intel based hardware and to try and grow their market share on Intel, than it is with Ultra5s and Ultra10s.

While it is true that hardware is scarce for 8-way Intel servers, Intel themself seem to be OEM'ing out the SC450NX (quad) and MS440GX (dual) Xeon motherboards (with chassis also). The SC450NX, while not cheap, offers a lot of power with the chassis included. Seems like little work for Sun to resell that type of hardware.

With this in mind, don't forget about the network client, that will continue to be important for many projects that could use Solaris x86 based servers (and not to mention laptops). If Sun can continue to penetrate into server, client, and laptop marketshare with Solaris x86, they could grow that section of their business into a very nice profit center. Keep in mind that for people using PC unix, there will be no development to single processor versions, Linux and FreeBSD will also be SMP kernels, and those PC unices will continue to run as clients and on laptops. Solaris x86 must support the Intel range of hardware better.

Things I want to see:

1. pkg formatted public utilities, compilers, etc that is kept up to date, something similar to SummerTime x86, but is given more care. SCO, DEC, SGI all ship their versions of "really useful utilities", I suppose Sun can do the same. Putting out a web site is not adequate. Who would waste 2 hours downloading egcs source code and build it themselves? Also, lower the price for the C Workshop products. No one in education will pay > $100 for a compiler.

2. Support for "major applications". We've been through this several times. First was netscape, then acroread, what now? Real Player, MacroMedia Flash, are the prime suspects now. We need native applications, not a Java one.

3. If Sun's purpose of giving away Solaris to education want to make sense, more support in terms of things that students will run is warranted. eg. ICQ, Video Conferencing like CUSeeMe etc will be most welcomed. Of course, an improved Sound card support is also needed. Also, of prime concern should be the "plug and play" support for non-postscript printers. Vividata is fine but too pricey. Configuring ghostscript is not the piece of cake for most people. Second is an easy to use PPP setup. This is also one of the weakest point of Solaris. The stock asppp is so difficult to setup for novice that this is the most frequently asked question on Solaris x86. (next would be running non-postscript printers.) Talking about PPP, a lot of people are running a network at home and would like to run things like ipfilter in their gateway box. Why can't Sun make up a "SOHO" server and throw all these stuff in? A bundle of Vividata's postshop, scanshop, faxshop and a easy to use PPP, with addins like NAT support will be very useful.

4. laptop support. I take my Tecra to lectures and show it to the kids :-) I think I'm very lucky in that I know what to install and what component to avoid (thanks to the mailing list). Otherwise it will be disaster. There should be a major brand of laptop vendor that is supported.

These two chipsets would make opengl run smoothly on Solaris x86 and could make Solaris x86 an excellent alternative OS to make opengl development on.

I would also like the SUN Xserver for x86 to support multihead. I do a lot of software development for the Solaris platform and I would like to be able to use my two monitors at home where I run Solaris x86. Also, I admire SUN's initiative of offering Solaris for free for people registered at the Developer Connection, but I also wish that they offered a cut-down version of SUN CC/C++ for free (or 100$ or so) even if it entailed having only access to the patches and not the regular support.

Also, I think that SUN should make deals with certain PC hardware manufacturers so that they would develop drivers for Solaris, and SUN would in turn certify and resell their hardware. For example, SUN could sell PC video cards from 2 or 3 companies that would make drivers for Solaris for their cards. It would eliminate a lot of trouble for SUN and increase the availability of drivers for Solaris x86.

I very much like Solaris to succeed on the x86 platform. However, this will not happen if a heavily used video card is not supported. Dell introduced this card into its systems in the summer of 1998. I believe 8+ months is more than enough time to develop the drivers for the nVidia 128 family.

I even considered replacing my current STB card with a Matrox G200. This is one of the most used video cards for PCs. Well guess what, it is not supported by Solaris (at least according to posts on the various Solaris newsgroups).

I just want to say that I have a couple Solaris Intel machines sitting around in my apartment, and I've really had little to no trouble with them. I've (fortunately) had around, or have been able to purchase affordably the recommended hardware on the HCL, and have had so, so few problems, that I forget how to use Micro$oft windows, etc. I've built all these suckers myself, and am super happy with the performance I get, even on lesser-powered boxes. I'm just amazed at what a great product Solaris x86 is, and really would be lost without it. Every now and then, I get a little snag here and there, but the great thing about most flavors of Unix is there's always a workaround -- and if that fails, as with anything in the computer world, you can throw money on the fire and just get some new parts and start from scratch -- which is nice on the intel platform which has affordable and various components at great prices. Someday I'll have the bucks to buy a nice Ultra workstation, but until then, my pentium and PII boxes will have to do. I am not a big fan of Linux (been there, done that with Slackware and Redhat), and while I think they've really done a great job with their operating system, I just don't like the feel of it, if nothing else (well, the die-hard loudmouths that jammer on and on all day about it bug me too). Anyway, enough of that. I've got two things on my wishlist for Solaris x86: built-in multi-monitor support, and perhaps a job at Sun this summer. I've already been to the site of the place that sells the product for x86 that will let you run dual-headed, but I really don't feel like shelling out a couple hundred clams for it, and I think it would be really neat to see in Solx86. I also sent in my resume for a summer job, and never heard back! =( Anyway, otherwise, I'll be using Sol/Intel until I can't see or type anymore, and have been an otherwise super-happy customer! (Actually, better backup utilities on Sparc and Intel would be nice, as well as some more printer support!). thanks. --andrew

We bought a Toshiba laptop specifically BECAUSE it was ON the hcl. When the supposedly supported system failed to function (ethernet card suggested wouldn't work), I was told by Sun that they were no longer supporting laptops. I am a strong advocate of Sun software, but it's very hard to fight the battle when you're getting knifed in the back on a regular basis.

If we had 2-D support; maybe, one day, 3-D?!

This is just an example of a chronic problem. I would love to recommend Solaris as a high end desktop OS, for both its stability and its simplicity compared to NT. But it's impossible to push the OS when it doesn't support _any_ current generation, high end hardware. Microsoft has a lot of developer/integrator support, but Solaris has to roll its own comparatively. Understanding that, I still can't see why I can't get a driver update for the hottest selling graphic controllers in history!

Same holds for other hardware. Solaris has to be sold as the choice for heavy duty needs, but it won't work if it doesn't support the hardware the heavy duty users have.

regards,

billr

We use a few notebooks as field data collection and reduction workstations (by field I mean without commercial AC power so we rely on generators and/or batteries). Our home grown apps were originally coded on Sun SPARCs using a combination of C, Fortran, and IDL (from Research Systems Inc). As SPARC notebooks have virtually no capabilities to run from batteries, we obtained a ThinkPad 755CX years ago and installed 2.5.1 and appropriate DUs to make it work. We later added an IBM 765D (not on any HCL but very similar to a 760 model on the HCL). We found that PCMCIA support was broken when we added more than 64 MB of RAM. As we really intended to use these machines in the field, we needed support so we bought a Sun Software only support contract for x86 server and software development kit and purchased several copies of the x86 Sun Workshop suite (C, Fortran, debugger, etc). When I called Sun for support on a pcmcia problem I was given the run around. Turns out that the problem was that the 2.6 FCS pcic driver failed on ThinkPads with more than 64 MB of RAM (our 765D had 104 MB). I asked on the x86 Solaris mailing list and a Sun engineer sent me a copy of a new driver built for the IBM 770 (early model) to test. That fixed the problem for 2.6 and the 765D. I later called Sun when I discovered that the 3C562 Ethernet/modem card from 3Com (it is on the 2.6 and Solaris 7 HCL) didn't work. I checked Sunsolve and found 4 bug IDs. Email from a different person at Sun indicated that they knew the fix but no patch had been built. After several calls I was told that there would be no patch for Solaris 2.6 but a fix would be included in Solaris 7.

When I installed Solaris 7 on our new 770E, I discovered several interesting things. First, the old PCMCIA problem for ThinkPads with the pcic driver remained. I tried the new 2.6 level driver I had for the 770 and that allowed the machine to configure cards like a 3C589D and several PCMCIA modems. However, the 770E will not boot if a card is plugged into the top slot at boot (it boots fine with a card in the bottom slot). Cards inserted in the top slot after boot are configured and work. I also discovered that Sun had not fixed the 3C562 driver. I wasn't real happy about that as I had been told to wait for the fix in the new version!

I sure wish that Sun had informed us sooner that they were dropping notebook support in x86. I see x86 on notebooks as a real plus for Solaris - we use SPARCs in the office but don't see a real way to use them in the field and saw x86 Solaris on high quality notebooks along with Sun compilers and commercial applications like RSI's IDL on x86 as a suitable field solution. However, to make them useful we need functional PCMCIA support as some of our field data collection devices store on static RAM cards. Also, we would like to be able to use PPP and/or ethernet on the notebooks and as Sun doesn't support docking stations (in general) we need PCMCIA ethernet and modem support. We need to transfer raw and reduced data from the notebooks to our Suns via the network back at the lab.

We would also like reliable docking station support as we could then use SCSI tape drives and PCI ethernet cards. I was able to use a docking station with the 755CX (Dock II) and the 765D (SelectaDock 2 or 3), however, the 770E under Solaris 7 (it will not install 2.6) finds the PCI ethernet card in the DCA but does not build devices (same for the Adaptec AIC SCSI chip and the PCMCIA slots in the dock) so no devices in the docking station are available to Solaris although I can use an external keyboard and mouse connected through the dock.

Sun also doesn't support the video chipsets used in modern notebooks. I have found a commercial solution for that in the Xi Graphics Laptop server and it works. However, it does not support DPS and XIL so it isn't a full solution like a real Xsun driver included in Solaris for various modern chipsets would be. I can, however, live without DPS as there are solutions but I would prefer a real solution.

I realize that we are attempting to use Solaris in a somewhat non-standard way, however, through Solaris 2.6, Sun seemed to think that what we were attempting to do was useful as they supported notebooks explicitly as they were on the HCL.

Sincerely,

Stuart F. Biggar Associate Research Professor Optical Sciences Center University of Arizona

I have evaluated Solaris2.6X86 against WinNT4.0 last year, the objective is to select OS for Store Controller in 1000+ of 7-Eleven stores in Thailand. Althrough Solaris is more functional and more reliable than NT but we decided to use NT because: 1)Solaris has fewer HW support. 2)Solaris has no Thai Support [a must] 3)Solaris has fewer app development tools. 4)Most IT staffs believe that NT is easier to use. Althrough I love Solaris but I have had a hard time defence people.

I am deploying a significant application which has been designed from the beginning to be run on Solaris x86. I require many low power cost effective computers that are able to be easily deployed and serviced. NT doesn't cut it from a reliability standpoint. Sun hardware doesn't cut it from a cost standpoint.

I could use Linux, but there is *tremendous* value from my customer's perspective in having a stable well regarded OS like Solaris. Having worked with Solaris Sparc since it was introduced and Solaris x86 for 3 years, I know that technically this OS can't be beat. The multiprocessor support, the advanced memory management and the most time-tested networking stack make Solaris (x86 and Sparc) the clear technical winner.

However, there is always a lingering doubt as to Sun's commitment. I see Sun field engineers running around with Solaris x86 on laptops, but no official laptop support. I see companies like Oracle offering Solaris 86 versions, but not on a Solaris Sparc derivative, but rather common with NCR MP-RAS and DG-UX. I also cannot buy Oracle 8i for Solaris x86 yet. Just today I discovered that Sun's very own product, Java Web Server 1.1.3 is only available for Sparc (oh, and to add insult...NT). The available list of commercial software is short, no doubt due to the same hesitation I feel about Sun's commitment to the Solaris x86 platform.

I see Solaris x86 as a golden egg for Sun. There are plenty of places where clearly Sun hardware is the best choice. However, there are many others where x86 based hardware is a perfect solution. I don't see this as cannibalizing Sparc sales, since they are different markets. As the x86 based system grows and requires the capabilities of Sun hardware, the migration is painless. This is much easier than trying to move an overloaded Linux or NT app to Sun hardware. Solaris x86 is the way to get all developers developing for Solaris. This has to be beneficial in mind share and more practically, a way to insure portability.

I agree that supporting the vast array of new hardware that pops up daily in the PC market is impossible. However, I would be satisfied with a small but leading edge set that is kept up to date. As an example, if Sun said that the only video boards supported were ATI, and the only SCSI boards supported were Adaptec, I could live with that. The flip side of this is that Sun would need to support each new ATI board as soon as it is released, and the same for Adaptec. I believe this policy could be extended to laptop support as well. Just pick a (respectable) vendor and keep those drivers up to date.

Another choice (although not mutually exclusive) is to publish existing driver code so that the developer community can build support where Sun is unwilling.

When calling Sun or Oracle, I always feels like the poor stepchild with Solaris x86. Technical support people are uninformed, and sales people are largely unaware of the product. This is a great product. Go tell people and really push it.

Ron

I travel around Asia-Pacific instructing our resellers/distributors how to implement/deploy our network security products. My preferred platform of choice is Solaris x86 because of its stability and its power over WindowsNT. However, I have no choice but to run WNT4 because of lack of support for Solaris x86 on the laptop (I have a Toshiba 460CDT).

I also wish that Sun would encourage more users to use the Solaris x86 as their workstation of choice by encouraging more popular developers to produce stuff for Solaris x86 (read: expand the Solaris x86 HCL !)

I've been working with Solaris long before it was Solaris and prayed that Sun would release Solaris on Intel. I got into that issue late starting with 2.5 and now using 2.7 here at work. The PCMCIA should be better supported. I would have love to use my 3C574 card rather than the 3C589D which still sucks cause the pic driver still doesn't work. Whomever gave me that pci driver showed some strength but really needs to work on it more cause I have to insert the PCMCIA card at boot time EVERYTIME. Give me a break Also on the Desktop version of Solaris I run on at home Could also see more application that would run concurrently with those Microsoft IDIOTS such as Excel and Powerpoint. Staroffice is ok but still has a long way to go to being fully compatible. I certainly didn't like being told that SUN isn't supporting LAPTOPS. Come on guys it is still INTEL what the BLOODY difference. I haven't tried the PCMCIA on sparcs in which I use heavily at work and for customers I support. I always get ask about INTEL version and I'm really sorry to have tell them as much as I like Solaris don't use it. If you do, Hardware depency is critical with Solaris. I still see that the market is playing to the hands at Microsoft wich totally SUCKS WIND. That is because vendors such as yourself is not pushing these other hardware vendors to support Solaris. I have to say right now, That Linux really sounds too good to be true. And I'll be damn to see Sun let that happen. LINUX in my opinion is ok from the stand point of small server but Solaris goes along way. BUT Solaris totally lacks anything on DESKTOP. WABI Suck then and Sucks NOW.

I'm planning on replacing about 25 Solaris x86 boxes this summer, and Sun currently supports none of the potential vendors' default video cards. I'd like to see better support for video in x86, especially the Diamond Stealth and ATI cards for AGP and some older PCI systems.

I came to Solarisx86 from Linux when I learned about the free promotion. I prefer Solaris now since I feel that it is more stable, and much better documented. Solaris is my desktop system now and I hardly use Windows any more. However, I think that it is important that driver support be uptodate. For example, I had to add XFree86 because the Matrox G200 is not supported by XSun. I also had to purchace PostShop from Vividata because my Epson Stylus 850 was not supported by the OS. At least, I found my own solution. A bigger problem is that Real Audio player 5.0 does not work on my Solaris 7. I see it as mandatory that Sun make every effort to offer "standard" WWW technology (RealAudio, RealVideo, Shockwave, Flash etc.)to the users of Solris x86. Fortunately the Adobe Acrobat is now available.

Another thing that I miss greatly is the use of my Connectix Coolor Quickcam for video conferencing. I think a (chip) videroconferencing option should be available for the platform (CuSeeMe!). Finally, it is mandatory that an MS Office compatible solution be available under Solaris. I use StarOffice which I like very much. The import filters for MS Office could be greatly improved however! There is great interest in Solarisx86 here and it will have a great future if drivers will be available for the standard hardware (video cards, soundcards, ethernet cards, printers, scanners, modems), Office compatible programs are supported and web browsing includes the most common offerings!

I would like to see improved PCMCIA support so that I can get Solaris on a laptop and network it. I use Solaris x86 for most of my work, and it would be a huge benefit to be able to use work in the same environment when I am on the road. The success i have had with Solaris x86 has opened the door for increased use of Sun hardware and software throughout our enterprise. Better hardware support, and a serious commitment from Sun which translates into more applications for x86 would encourage me to recommend Solaris x86 to clients for a variety of applications.

I could have purchased almost any notebook but ended up with this one since all the ones on your HCL were out- dated and nolonger sold. I have used other companies software to run Xwin and sound. Also my network adapter the Xircom pocket ethernet parallel adapter does not work with this computer. I would like to see your driver updated to continue to support this adapter with faster computers. Thank you very much.

Sun needs to continue support for PCMCIA as well as continue to support laptops. The current pcic driver doesn't seem to work properly and has no "user configurable options" per the man pages. I will not be using Solaris 7 on my laptop and will be looking for a complete client/server solution to reccomend to my Corporation's clients. Solaris makes a great server platform, but without complete client/server support it falls short of the competition in the real world. It seems that even Sun users are currently using Windows 98 on desktop systems inside Sun, and the Sun employees are most likely running Windows 98 as they do on their desktop, or Linux, as reported by some Sun employees. With the addition of Sun backing Linux, it could be that Linux is a better solution for client/server (i.e., Both!) on Intel hardware.

I still see I/O as a bottleneck on my system, I hope it makes it in the next DU for 2.7!