Slashdot Mirror
Sun's Linux Exec Departs
HyperbolicParabaloid writes "The NY Times (free reg blah blah) has an article about the departure of Sun's no.2 exec, but also mentions that
Stephen DeWitt, the vice president of an important business unit that leads Sun's efforts with the Linux operating system, quietly left Sun
on Friday, the company confirmed today."" And the question is: How will this affect projects like OpenOffice release and the on-again, off-again McNealy Linux relationship.
I think this is worse news for Sun than for Linux.
There is little or nothing Sun's OS can do now that Linux can't. And if Linux can't now, it will do soon.
I know that MS haters like to see Sun as in some way a friend - my enemy's enemies and all that - but the logic of a free OS applies just as much to Sun's offerings as it does to Microsoft's - maybe even more so as what Unix application will run on Solaris and not on Linux.
Sun will sooner or later have to realise that Linux will dominate the Unix OS market to an ever greater extent in the future. They will not have much oa future if they don't factor that into their plans.
Believe it or not, Sun IS good for Linux.
Sun has quietly been giving some assistance to the "corporate" Linux developers. Sun makes its money on hardware, and witha majority of the younger generation coming into the job market with a strong knowledge of Linux, Sun is smart enough to know that their OS (which they don't make money from) has its days numbered.
Sun's best bet is that they gradually convert Solaris to a more Linux-like system if not going Linux entirely.
Why do you think they were going to drop Solaris x86? Why do you think they didn;t complain when people started porting the Linux OS to the Sparc platform? Why do you think Sun GAVE machines to people porting Linux to Sparc brand-new systems to continue their work? Why do you think since Solaris 2.6, the Sun OS has started to go back to its BSD roots and take on more of the Linux characteristics? Why do you think that with Solaris 3.0 the OS is going to have a more RPM-like system of patches and installs instead of the antiquated PKGS?
Think about it...
As someone who recently had some admin work to do on a Cobalt RaQ3 I can only testify to the bone headedness of SUN's Linux efforts. I have the following points to make about SUN in relation to their so called Linux efforts:
.pkg installer. Considering that the system is based on the RPM system one does wonder why they didn't go a more compatible route.
It is nigh on impossible to get updates for third party packages (.pkg's) for older Cobalt machines. Cobalt had the brilliant idea of making a web browser based admin interface, thereby supposedly making it easy for newbies to administer the machine. This has the tangential effect of making the machines vulnerable to cracking (like IIS) because the admins have no idea of what they're doing and what they should be updating. Turning the machines over to someone who knows what bash is doesn't help immediately because installing software from the commandline is difficult as the whole system has been modified by SUN to make it difficult to get those CLI installs reflected in the web interface.
The news groups and online Cobalt boards of full of irate users asking for help, and , more importantly, not getting much from SUN. Almost all help is from other users. It took me almost three days of constant searching to find SUN documentation on how to roll my own
I needed a PHP update and a custom Webalizer in German. The PHP "make install" exited with an APXS error and after about a week someone told me that the Cobalt APXS script is buggy and outdated, after which I managed to do the install by hand. The Webalizer compilation was less error prone but the fact that I had to do it and the PHP installation by hand because there were no packages available says legions about SUN's commitment to the platform.
To get help from SUN you have to pay, and considering that you already payed for the machine and ISP costs etc, it is a slap in the face.
The experience was frustrating and only strangthened my conviction that SUN has almost no idea of what consumers and smaller operators want, and possibly that SUN will go out of the market because of this if they carry on in this manner.
Sun is facing an inexorable onslaught.
If they thought "Wintel" was creeping up their food chain, "Lintel" is lower priced still and hungrier for the UNIX market.
Sun needs to capitalize on its UNIX experience and to become part of the Linux solution, rather than reactively viewing circumstances as the Linux problem. They've made some good moves already, in terms of StarOffice acquisition and having some developers work on Gnome. But they need a coordinated vision that puts everything together. E15K database ervers working well with Linux server appliances which interact well through all the built-up Unix infrastructure (NFS, etc.)
IBM and HP have already seen the handwriting on the wall and are doing things to take advantage of the shifts going on in the marketplace.
Sun certainly has a lot to offer, they should put someone in charge who knows how to leverage that UNIX experience and to grow new markets based on their existing network of sales staff.
Java could figure prominently in such a strategy; but promoting Linux on SPARC seems to be more of an uphill battle, AFAICT.
"Provided by the management for your protection."
...they'd embrace the upcoming AMD Hammer architecture and build high-end Linux/Hammer workstations and entry- to mid-level servers around them. Kind of a Sun/Linux meets Apple deal: clean product line, not a lot of overlap, well designed, very cool. I say Linux rather than Solaris if only because it's impossible to do Solaris/x86 device drivers for everything, whereas Linux has a decent chance.
Of course, that'd freak out their SPARC people, so they'll never do it. Pity.
They may have lost their Linux Exec, but they recently acquired Whitfield Diffie! (For those not aware, Whitfield Diffie is one of the inventors of public-key cryptography, the technology used in PGP and elsewhere)
Alex Bischoff
HTML/CSS coder for hire
"Sun goes through executives faster than french pastries at a Weighwatcher's convention."
... funny?
Huh, a tired form of a joke that wanted to be said so the first company that came to mind was popped in. They still have the same CEO. Same chief scientist. Matter of fact, the top still looks a lot like it did just out of Stanford.
Don't let that stand in the way of being
I've finally had it: until slashdot gets article moderation, I am not coming back.
I too when given the choice between Solaris and RedHat would go with Red Hat, except I would be uncomfortable about it.
When I pay more for the Sun Solaris system I know I can also pay (alot of money, yes) for support from Sun using their Sun Spectrum support. Their support is great. If there is anything wrong they will fix it. I don't use their support that often but when something comes up and I call them and they do.
I don't know that the support for the RedHat box would be as good. I know the support from the Dell, the hardware vendor that my agency uses for thier x86 servers, is not as good as Sun by far.
An example: I call up Sun and am attempting to find out why I'm getting some errors on my raid array. After a little troubleshooting with a technicition they have me run a utility which gathers all the relavant information about my system and I ftp it to them. (I don't remember the name of the utility as its been a long while since this happened). They call me back in 10 minutes and tell me that it has to do with the OBP firmware level. I upgrade and it works.
With Dell, my latest example is a bit more recent. We have a 130T tape library. Its one of those four tape drives 1TB DLT jobbies. We have three other L1000 units in various places (and ALT library relabeled and supported by Sun as their own) but we have this one 130T library connected to two Dell servers (which are running windows, even more unfortunate). Its not working. One Dell server is connected to the robitic controller and two drives and the other server is connected to the other controller. We call Dell and they have a guy (from Unisys who subcontracts for Dell repair in Portaland Oregon) who comes out and they start replaceing stuff. They replace virtually everything, including the controller board, on the 130T. We call up Dell support while we are doing this. This is their 24x7 gold-super-important-system support line. They get wind that the 130T is connected two not one server, but two. That's it. We aren't supported. The system has worked since we installed it a year ago, and aside from a few server related issues, the library hasn't been an issue, and there certainly isn't a problem doing the same thing with any of the L1000 units. But once the technician heard that our library was connected to TWO servers, we where in an unsupported configuration. Our Techie on the phone with them was so flabbergasted that they would do that he didn't even argue, he just hung up the phone and got the library kinda working with the two tape drives he could get working.
So if I could be sure that I can get Sun Spectrum support for my linux box, even at a premium price, I would go with them. They could proclaim that they are selling Linux boxes, and they have ported Linux to Sparc or they have started using 64 bit x86 chips in a new line of servers that run Linux and they have opensourced the Solaris kernel and have people working on kernel patches for Linus. These are all just idea's which their bean counters don't see the botom line on but would get people like me announcing they are going to join the church of Sun. They could be to the unix system what Apple is to the desktop system. Well they do have a bit of that going on already with their expensive cute purple boxes.
In the mean time Sun will have a product to sell but I suspect they will be suffering from hypoxia if they don't change something quick. I hope they do, because I do like their stuff.
Their 880R servers are prices quite reasonably when configured with 8 CPU's, compared to a DELL box.
You know, when I read comments in the Linux fora (like Slashdot, for example) that imply or explicitly declare Linux superior to Solaris, I have the impression that not many Linux evangelists have really administered Solaris. I don't mean just telnet to a Solaris box and run IRC or compile something.
... and trying to keep a positive cash flow. Which isn't an easy market.
The problem isn't "Isn't Solaris sometimes better?", it's "How large a market do we have where Solaris is substantially better? And how much will people pay?".
I've heard people complain about the GUI, the useability, the friendliness, etc. of Solaris. But I haven't heard anyone (except, perhpaps, HP or IBM) say that it didn't handle large environments better. But the "large environments" is a relatively small market. And it currently has IBM, HP, and Sun (plus others?) already in it. Well, IBM has decided that it can scale Linux from the bottom to the top, and therefore gain a different kind of advantage. And that it can handle the large machine problem by running several different sessions of Linux. I'm not clear what HP is going to do, and I'm not sure that they are either. Except for merge with Compaq/Dec. So HP is currently distracted, but when they come back they'll have some new hardware behind them, and another version of Unix (so they'll be supporting HPUX, Linux, and, perhaps, Dec Unix (well, maybe not). IBM is saving on development costs, and, additionally, tuning the commodity OS (Linux) in directions that favor it's hardware. (Consider all of the contributions that IBM has been making recently. I would bet they generally have the effect of making Linux run better on some piece of hardware that IBM sells. Would you bet against that?)
So Sun is needing to pay for the development of Solaris anyway. They can't drop that without greatly offending a huge number of customers. And if they want Linux to run better on Sun hardware, they need to do something to cause it to be tuned for that, but what? OpenOffice was desireable because it was both good for Solaris systems and bad for MS Office. And it got them a lot of favorable PR among developers. But anything that they put in on Linux is to the detriment of Solaris. And Solaris is their proprietary system. Which they have to keep paying for anyway. But if they don't support Linux, then it won't run as well on their hardware. And it's the system that an increasingly large number of people are already familiar with. So they would loose that advantage.
Ugh!
If their hardware was sufficiently much better than the rest, then they could just switch most of their development over to making Linux run better on their hardware. As Solaris died back to a bug-fix mode their expenses would decline (it costs less to help Linux run well on your software than it does to develop a complete OS). But this would mean the loss of their captive market. (Everyone could suddenly move from Sun to, say, IBM with no conversion costs to speak of.) Not desireable (for Sun).
Oops!
I don't see any good way out for them. Only by concentrating on making better hardware cheaper
I think we've pushed this "anyone can grow up to be president" thing too far.
Background:
;) so believe or disbelieve the following at your whim.
...
... but I would be surprised to see another major Qube product (ie, new hardware) for another 8-16 months.
I worked for Cobalt from before they went public until about 6 months after the merger. I moved away from the Cobalt business but I still maintain close ties with people who work directly for the Cobalt business unit within Sun. To keep anyone from getting in trouble, I'll stay anonymous (too bad, the karma would have been nice
Perspective on Mr. Dewitt:
* Stephen Dewitt did tremendous things for Cobalt when they were their own company. He helped bring them public in a big way and had the vision to sell to Sun for a large amount just before Cobalt's main markets (ISPs and small VARs) tanked in early 2001.
* However, once the merger happened, Mr. Dewitt was moved a couple of steps above the Cobalt business unit to lead Edge computing. He was in a much more visionary position than he was with Cobalt and had very little to do with day to day operations. Most of the Cobalt and Linux folk within Sun had heard nothing from him for months before he resigned as he was focused more on edge computing strategies than specific products.
* It's quite possible he was just being graceful and staying around long enough to make sure that the Cobalt unit got incorporated as smoothly as possible before he went on to find the next company to lead. Mr. Dewitt is not one to easily accept going from being a CEO to being a vice president, even of a very large company.
Perspective on Cobalt:
* Between the market collapses and a very complicated merger, Cobalt products have suffered. They are probably a year behind where they would be if neither event had happened. However, if they hadn't been bought by Sun, they also could have easily disappeared when the bubble popped, so overall it was probably a wise move.
As for Cobalt products
* The Qube product got some pretty major feature additions back in February which were given out free to anyone with an older Qube 3. Not bad
* There have been allusions to a new RaQ-like product that will probably be released sometime this month that will be on par with the appliances that IBM announced last week based on the X series and Sphera software. I don't have much more details, but expect a refresh of that product line soon.
* Expect to see a new product from Sun by the end of summer that is a general purpose box similar to the 1U boxes from IBM/Compaq/etc. No real details here but I know that the Cobalt unit is the part of Sun that is running with this project (at least as of 6 weeks ago). It doesn't sound like Sun has alot of other resources working on Linux (except for their Blade project, which is a while off in the distance) but hopefully if this product line takes off it will cement Cobalt's position within Sun.
Heck, I might even try to go back if they look to be stable after the general purpose box, I jumped ship when the rumors of lay offs (which happened a couple months later) got too strong.
What if you took Red Hat Linux 7.2, removed the Linux kernel, and substituted the Solaris kernel?
On x86, you would immediately have 8-cpu scalability with no problem, with much greater efficiency than RH Advanced Server.
However, if you needed greater than 8-cpu scalability, porting Red Hat/Solaris to an e15k would give you a max of 106 cpus in a system architecture originally designed by Cray.
In other words, Sun could get control of the low end, provide a market-wide migration path to SPARC, and cut the throat of any high-end UNIX on Intel. They would instantly own the "Linux momentum."
They need to do it now. Right now.