LWN: Comments on "Scott McNealy steps down as Sun CEO (Mercury News)"

OT: third world salaries

man_ls — Tue, 02 May 2006 21:56:27 +0000

Thanks for the info. Here in Spain it's hard to get 30k programming in COBOL; it is certainly possible programming in Java, but not much more. Aren't we cheap.

Oops

X-Nc — Tue, 02 May 2006 21:43:25 +0000

> I must confess my deepest ignorance about COBOL. The things I
> know are: everything is done on a green screen, many things
> are done in batch processing, and it's horribly paid.

It's all done on nice, shiny GUI screens now, most stuff is done in batch (which is much preferred for doing massive data manipulation anyway) and the pay is quite variable. Back in the Doom-and-Gloom 90's with the whole nonexistent Y2K Bug crap you could get upwards of $250K to do COBOL stuff. Now I think that it's more in the $70K to $150K range.

> Did I mention that there is no libre software in COBOL listed on FreshMeat?

No, but I would have mentioned it myself, too. You'll not find anything dealing with COBOL in the FLOSS world. First, there's not a big need for it in this community and second because no one in the community has a clue what the language can do.

As for your examples, they are a non sequitur, especially the "<>" example. COBOL was designed and built to be read as it's own documentation. As such, using actual English words was preferred to cryptic character strings. Adding things like "=" for EQUALS and ">" for GREATER THAN is purely for ornamental purposes.

If you remember that programming languages are tools then you'll be better able to see the value of any particular language. You wouldn't use COBOL to write an OS but you certainly wouldn't use C/C++ to write something like the banking system or airline reservation systems. The right tool for the right job, be it Java or C or perl or COBOL.

Oops

man_ls — Tue, 02 May 2006 21:16:23 +0000

I must confess my deepest ignorance about COBOL. The things I know are: everything is done on a green screen, many things are done in batch processing, and it's horribly paid. Maybe COBOL here in Spain is outdated, but it certainly doesn't look like a modern language. Did I mention that there is no libre software in COBOL listed on FreshMeat? (I did find this "open-source COBOL compiler", but it translates the COBOL code to C so it's not really a compiler.)

Maybe you could give us a pointer so we can check it out. I can only find this link where it says that the projected new features for COBOL are things like: "dynamic tables, increased size limit on non-numeric literals from 160 to 8191 characters, allow <> As Synonym for NOT EQUAL TO". Hardly modern stuff.

Whoa there, Nelly

X-Nc — Tue, 02 May 2006 18:37:20 +0000

> [Java] is, as a lot of people have pointed out,
> a modern COBOL, hard to leverage for anything;

Funny you should put it that way. The only people who would consider COBOL to be anything but one of the most modern languages are those who know nothing about it. COBOL is more object orientated than Java or C++. It's an advanced and modern language, just look at the specs.

As for "leverage", COBOL is by far the best language for manipulating and handling data that there is. The only problem that COBOL has is an image problem. Everyone thinks of old code from the 60's and 70's when they hear the word COBOL. Nothing could be further from the truth. If TPTB were to actually keep the IT/IS world knowledgeable about the language it might not get such an undeserved rap.

Sun to follow SGI

man_ls — Tue, 02 May 2006 16:20:05 +0000

What do *you* think they should do, to be profitable?

Let's see... sell some stuff? Like in make boxes people want to buy? Right now Sun have to keep an operating system and support a lot of software. At the same time they make poor processors: even their partner, Fujitsu, beats them in price and performance. Now that their hardware looks more and more like standard x86 stuff, why spend more?

Sun to follow SGI

man_ls — Tue, 02 May 2006 15:56:33 +0000

the installation of Linux reflects more on a desire to use x86 hardware compute performance than anything in Linux per se.

If this was true, Linux for UltraSPARC would have no meaning and no users.

I'm just suggesting that they could try and use Java aggressively to achieve it.

The "Java brand" carries little weight in the Linux world, and little more in the Windows world. It is, as a lot of people have pointed out, a modern COBOL, hard to leverage for anything; abandoning Java for Linux might keep Linux out of some markets for some time, but not for long. If Java is their last chance, as you suggest, then you might as well sell your stock.

If Linux disappeared tomorrow, would you care?

Of course. However, if Solaris disappeared tomorrow, I couldn't care less. I guess most of their users would just install Linux and keep going. People often quote some features which are unique to Solaris, but most of them (containers, ZFS...) are new in Solaris 10, so few people will rely on them yet. And Linux is catching up fast anyway; pour some resources in this direction and watch it grow.

Sun to follow SGI

man_ls — Tue, 02 May 2006 12:54:26 +0000

Sun publish free software: yes. But why? If it becomes too hard to make revenue, why would they continue?

Because they make their money selling hardware? And because they want software that runs flawlessly on their machines. And because they need to give their customers what they want.

You say 'Niche, specialty markets are surrending (sic) all the time' but I don't see ATI and nVidia surrendering, and they are hardly niche players. What's your point?

My point is that niche, specialty markets are surrendering all the time. ATI and nVidia are not niche players, so they are not surrendering (or if they do, it has nothing to do with my statement). Let us reformulate it: closed hardware providers tend to disappear while open alternatives flourish. ATI and nVidia are not in that game; they follow a set of public specifications (like VGA, OpenGL and many others) and other companies can play and compete with them, like Intel, SiS... so ATI and nVidia have nothing to do with our little problem.

Hardware providers all tend to go the standard route. If Sun is going to go into the business of making hardware only they can use, like the TCP/IP offload engines you mentioned, then they will have less chances to survive, not more.

ATI and nVidia support is not as important as you picture it, either for Linux or for Solaris. For Linux, if you want to have binary drivers you can install them; they are only really necessary to play games. For Solaris, there is hardly a market on the desktop because nobody needs it. JDS is just an example of why a Solaris desktop is a "pointless foray", as you put it, and ATI and nVidia support is definitely not needed for (in your words) "running Sybase and Oracle and Java server applications". So we may conclude (again) that your mentioning ATI and nVidia was a red herring. Sigh.

Sun follows DEC

pyellman — Mon, 01 May 2006 17:56:37 +0000

Now, that's just crazy talk. That would be about as clever as Sun slitting its own wrists to see if it could drown linux in blood.

Peter Yellman

Sun to follow SGI

jmansion — Fri, 28 Apr 2006 14:00:44 +0000

> you confused IBM with Sun

What was confusing? If Sun withdraw their Linux JVM, customers wanting a real JVM can turn to IBM. But I don't think we would except in the short term, even though the IBM JVM is pretty good.

> Linux with Solaris

Well, I'm confused by you! No, Linux and Solaris are both operating systems that can run on x86 in 32 and 64 bit mode, and also which run on SPARC hardware.

Sun publish free software: yes. But why? If it becomes too hard to make revenue, why would they continue?

You say 'Niche, specialty markets are surrending (sic) all the time' but I don't see ATI and nVidia surrendering, and they are hardly niche players. What's your point?

If Sun make Solaris on x86 work better with closed source ATI and nVidia drivers than the Linux kernel developers *are prepared to* then why would Solaris not be *potentially* attractive as a kernel while retaining the same software stack cf Nexenta? (I'm assuming that they fix the other hardware issues they face on generic kit too, of course)

Linux does not have much volume on the desktop. In order to win it against Apple and Microsoft, does it need top notch support for the grahics cards people actually have?

What do you think will happen?
a) nVidia or ATI opens up their technology (and the other follows)
b) the kernel devs give in and stabilise to help nVidia and ATI deliver closed technology
c) Sun catch up to a usable extent with Solaris11 support for crappy whitebox hardware, which has been woeful, *and* they leverage their customer relationship with ATI and nVidia for the Ultra 20 workstation.
d) open graphics hardware magically evolves to compete with ATI and nVidia
e) none of the above: Apple and Microsoft are sitting pretty

The question really is: could Sun achieve c)? If they do, is there a market for the result *on the desktop*? If not, is there a market for any other UNIX on the desktop, except where its sold specifically to run on Apple hardware?

If you don't think Sun will achieve c) because they've shown monumental inability to deliver that sort of thing before, then I'd have to agree with you. I'd like them to though, because I'd like a nice solid UNIX environment with GUI bells and whistles out of the box on my desk without having to pay the Apple Tax, or bothering myself with the ongoing open/closed driver debate.

I don't know why you mention JDS. Who cares? Sun is important for running Sybase and Oracle and Java server applications in financial institutions, not JDS and other pointless forays.

Sun to follow SGI

jmansion — Fri, 28 Apr 2006 13:32:35 +0000

> Have you ever stopped to wonder WHY Solaris on PCs was nicknamed SLOW-ARIS

Why wonder? I did try 2.5.1 on x86 when it came out but I'd rather run XP and a variety of other systems, so I do.

Have you actually tried developing and running non-trivial systems with gcj? Know anyone that does, in a production environment? Its getting better, but in Sun's target market it doesn't seem to have traction yet.

Ask yourself: *why* do you care about Linux? Its just an operating system.

Do you really think that the rather large number of businesses that are developing and deploying with Java are really about to move away from Sun's JVM? Even to IBM's JVM? A lot of these businesses develop and deploy on Win32 and need that portaability. Which open source JVMs will address that?

OK, Harmony will. And SableVM is a leg-up for Harmony. But its not really usable yet.

I tried to make the point that Sun *could* try to use the success of Java as a lever to get the shiny new Solaris on x86 into their target customer base. I would if I was them. But I'm not them, and I don't think Schwarz will do it either.

What do *you* think they should do, to be profitable?

This isn't about whether someone *should* hurt Linux, or whether I (or anyone else) wants to, but about how Sun's management might make returns for the shareholders, which is its duty.

> up the butt of something

Have you ever considered that Linux is not the only answer to every question, and people who use other systems (even proprietary ones) sometimes do so for entirely rational reasons? Don't you worry that fanboys and evangelists are fools to put the answer before the question - even when they are pushing Linux? Is everyone who even considers that Linux is not the Holy Grail 'up the butt'?

Sun to follow SGI

jmansion — Fri, 28 Apr 2006 13:17:06 +0000

> bottling up that linux hatred

Why do you think that? Its true that I think its just another operating system and just as crap as the rest of 'em, but I have been using it since the first Yggdrassil bible came out and I used to get SLS on a subscription of floppies - and my central heating system is run by an old HP Vectra running SUSE.

I think you are right that Linux is causing Sun pain. But in the datacentres in the City there's a lot of Solaris, and the installation of Linux reflects more on a desire to use x86 hardware compute performance than anything in Linux per se.

The issue here is: what can Sun do to stop it? Sun have tried to embrace Linux and be nice guys, but how's that going to help them be profitable? In my viewe Sun *need* to get Solaris10/11 into the space that Linux is invading in their traditional marketplace. I'm just suggesting that they could try and use Java aggressively to achieve it. If they don't, its clear that Linux on Dell will eat their breakfast - and their lunch.

Ignore ISPs and shared hosting and that stuff. If you're deploying Java on x86 you're either in W2k or Linux now, in all probability. Ignoring the Win2k deployments, which usually have good reasons to be that way, is the important bit Java or Linux? Would you be inconvenienced by moving to Java on Solaris on the same hardware? Heck, even FreeBSD has *finally* got a JRE.

If Linux disappeared tomorrow, would you care? I wouldn't, really. Linux has invigorated UNIX-based software development, but *BSD and Solaris would remain, and they're all 'good enough' for most tasks that one might consider Linux for. Red Hat *is* important and you clearly underestimate that, at least if you regard displacement of existing UNIX and Win32 servers in business datacentres as important.

James

Sun to follow SGI

petegn — Wed, 26 Apr 2006 23:10:53 +0000

Have you ever stopped to wonder WHY Solaris on PCs was nicknamed

SLOW-ARIS or are you too sheltered and up the butt of something to have noticed and as for Java well i doubt it's continued longevity it will no doubr be replaced in the not too distant future with an OPEN SOURCE replacment of that you can be fairly certain ...

Pete .

Sun to follow SGI

man_ls — Wed, 26 Apr 2006 21:22:13 +0000

I'm sorry but your message is too confusing. It is hard to follow because in a few key places you confused IBM with Sun and Linux with Solaris, or something.

Sun is a very important publisher of free software, and they would be embraced happily by the community if only they chose to play nice all the time. But then some company executives open their big mouths and lose all that love in a second. Schwartz is the worst of them all, and now he's the CEO. Pity.

Right - all that puff about nVidia and ATI chipsets is just a figment of my imagination. Same with assorted RAID controllers and TCP/IP offload engines.

Open architectures have public specifications and compete on efficiency, price and all the rest. Graphic cards support OpenGL and the Microsoft equivalents for a reason; otherwise you would have games that play only on one brand but not on the rest. Same with your other examples. Niche, specialty markets are surrending all the time. If you choose not to see it then it's your call. But Sun knows their fortunes are reversing.

You know, Red Hat makes 1/10th of Sun's revenues, but without moving a screwdriver. I agree with you that Sun might sell a unique mix of hardware and software; instead they market their absurd "Java desktop" based on Linux (recently Solaris) but with proprietary "Java" additions, as if anyone cared about the "Java" brand. When they don't sell any they wonder "what did we do wrong".

Sun to follow SGI

pyellman — Wed, 26 Apr 2006 21:05:53 +0000

Wow jmansion, sounds like you've been bottling up that linux hatred for quite a while! Must feel good to let off some steam!

>I think it has to try to hurt Linux to do that, if only in the short term, while it gets its act together over the ISV and hardware situation.

Umm, I've got some bad news for you: the potential for linux to cause Sun "pain" far exceeds the other. Why frame it in those terms?

Red Hat, Red Hat, Red Hat, Red Hat. At least you've got a name for your anger. Unfortunately, Red Hat does not equal Linux. How long has it been since you took a peek outside of the Sun marketing tent? 20 years you say?

Peter Yellman

And then there were two...

iabervon — Wed, 26 Apr 2006 16:06:04 +0000

Scott's keeping the same job title that Bill has, so I wouldn't count him out of the running quite yet.

Sun to follow SGI

jmansion — Wed, 26 Apr 2006 15:11:11 +0000

> It would leave IBM as the sole provider of Java for 99% of the market

Not so - it would leave Sun supplying to 99% of Linux, but the market for Java is mostly Win32, with *NIX variants following.

I don't think IBM could take over the Java process and advance it, not least because they'd have to work very hard to avoid Sun rolling on with Win32, Solaris and Apple support.

> specialized hardware is losing ground all the time in favor of open architectures

Right - all that puff about nVidia and ATI chipsets is just a figment of my imagination. Same with assorted RAID controllers and TCP/IP offload engines.

I'm not saying that Sun won't suffer (not least in an about face) but we're discussing what they can do to avoid becoming Yet Another Box Vendor.

As I said, I think the way forward for this is to create hardware that has as much proprietary acceleration as is reasonably possible to differentiate it from the likes of Dell. I doubt they can beat Dell at its own game.

Given that Sun will want to protect it existing SPARC user base (and in the business centre I work in, there are a lot more of those than there are Linux users, at least for critical business applications) then it needs the Solaris brand to remain strong rather than cede to running Linux on the SPARC hardware - even though Linux does seem to work pretty well on it - simply so it can control the opsys/hardware interface and its intellectual property.

And in order to do *that*, it needs to get traction for Solaris on PCs so it can get the departmental servers and racks of HPC blades all on Solaris along with the database servers and big iron - which would be an admin and support win for customers. And ISVs. But not Linux vendors.

I think it has to try to hurt Linux to do that, if only in the short term, while it gets its act together over the ISV and hardware situation.

I don't doubt that Schwartz will *not* do any of this, because he's nailed himself to the open source tiger rather firmly: in my view, before working out how Sun will make money.

Personally I hope Solaris *does* get a lot of datacentre custom that's currently going towards Linux because a) the competition is healthy and b) I'd personally prefer the hardware+software solution that Sun can provide and c) I've been developing and delivering on Solaris for 20 years and have been pretty happy overall with stability and the longevity of installed systems. Sun's binary compatibility record is pretty good.

Most of the business-oriented stack that runs on Linux will run on Solaris too, and as a user I probably don't care about the differences. But you have to ask yourself whether Sun or, say, Red Hat have a better proposition for support in the enterprise. Sun has more experience. And Sun has more control over its product.

Sun to follow SGI

man_ls — Wed, 26 Apr 2006 09:04:36 +0000

That is not a very sound strategy, IMHO. It would leave IBM as the sole provider of Java for 99% of the market, relegating Sun to an even more niche role. They would lose any credibility to advance the specification.

As to the rest, specialized hardware is losing ground all the time in favor of open architectures. Non-mainstream manufacturers are being relegated to embedded systems, and even there they are being replaced by cheap x86 (or even PowerPC) systems. You may not care about open drivers, but most people do, in fact Sun realized it had to free the source code or perish. It seems you would like Sun to go back to a closed platform with closed drivers; in essence it's like you want the 90's to come back.

Sun follows DEC

jmansion — Wed, 26 Apr 2006 08:31:31 +0000

> dropped Solaris for Linux Sun would be in a much better position now

How, precisely?

Personally I think Sun should drop Linux for Solaris. Gloves off, and do what they can to move Java-on-Linux users to Java-on-Solaris users, and if that means rocking the boat and terminating licencing for Java technology except on Solaris and Windows, then so be it.

Sun needs Solaris to (be able to) differentiate its hardware, especially if it moves to hardware that's much more like white-box AMD, so it can build in accelerators etc that have closed drivers. As a user I don't give a stuff about drivers being open, I just want the ethernet chip, raid chip and GPUs to run at full speed.

Sun follows DEC

X-Nc — Wed, 26 Apr 2006 05:27:49 +0000

Maybe now Sun will realize it's a HW vendor and stop with all the software/OS nonsense. No, it's to late for that. They's already let "Intel inside" so they are s-k-rude. You know, ten years ago if Sun had spent it's resources making top-end HW and also making affordable workstations and dropped Solaris for Linux Sun would be in a much better position now. First the Alpha, now the Sparc.

Scott McNealy steps down as Sun CEO (Mercury News)

richo123 — Wed, 26 Apr 2006 02:43:48 +0000

Been a bad week for Scotland. First Mclellan and now McNeally.

Scott McNealy steps down

b7j0c — Tue, 25 Apr 2006 23:10:53 +0000

agreed. if schwartz hopes to make sun relevant again, he needs to come out swinging with a series of dramatic changes, if only to goose the stock.

step one is to open up java. the time to debate this has long passed, java has gone from being a transformative force in development to simply being today's cobol. don't tell me sun "needs" to "control" java...java has ALREADY fragmented (jdk,jre,eclipse toolchain,gcj,etc,etc), i can't imagine the java world being in worse shape.

its also not clear if sun should be doing cpus at all. regardless of their technical merits, sun processors been driven out of the market. and to be fair i would say its probably time for a large chunk of the workforce to go. better now while the bay area economy is vibrant and they can get work elsewhere, and also while sun still has the cash to do a meaningful buyout.

cutting alone won't do it though, sun needs to come out swinging with something paying customers can run with. their low-power/green computing initiatives are a good start, this is something more server buyers are going to be forced to deal with. the power bills for large server installations are absolutely insane.

far too late

b7j0c — Tue, 25 Apr 2006 23:00:54 +0000

at least five years too late to save sun. what is the way out for this firm at this point? while obsessing over bill gates, scott missed commodity computing. EXACT same mistake larry ellison has made. i give sun five years, if only due to their cash horde.

Scott McNealy steps down

alspnost — Tue, 25 Apr 2006 21:47:26 +0000

It's hardly a dramatic change, but poor old Scott did need to let go. Sun has been directionless for some time, and this once great company is still floundering, with its long-term survival at risk. Sadly, I don't think Schwartz is exactly the guy to turn things around, as he's still old blood. Sun needs a big kick from some new blood, otherwise much of their top class technology will languish and go to waste.

And then there were two...

tjc — Tue, 25 Apr 2006 21:41:24 +0000

Well, that narrows the field a bit, leaving just Bill and Larry in contention for "Megalomaniac of the Year."

Scott McNealy steps down as Sun CEO (Mercury News)

horen — Tue, 25 Apr 2006 20:11:36 +0000

Shouldn't that be "Schwartz"?

May the Schwartz be with you!