Let's start with the good. Sun did something really quite amazing, it managed to eloquently explain what a grid is, and why you should care. It basically is a server bank that you can hire by the CPU hour. You can also buy storage by the GB/month. Both, not coincidently cost a dollar. You need 1024 CPUs for a day, 24576 dollars, no muss, no fuss. It is aimed at things that you need a cluster for, but don't want the overhead. If you ever wondered how Niagara and Rock will fill all those threads, your answer just arrived.
Sun's aim here is to change the way computing power is bought, and reap the economies of scale with a commodity product. There is a price for a service, no maintenance. Sun kept comparing it to electricity and occasionally to water, you don't care about where it comes from, just that is comes on when you plug something in. The price is either fixed, or competitive, and you use what you want, when you want, how you want.
Sun is the first to admit that the idea won't work for everything, and most pundits will point out that it may not work for anything. I think it is more toward the Sun outlook, if it can pull it off technically. There is a large class of work that will fit nicely into the Sun model, like render farms and oil and gas exploration, many of which would pay for more power on tap. Get it, tap? Har har, I kill me.
Let's assume for a minute that it worked out the technical hurdles and it will do what it promises. Let's also assume that it has customers lined up to buy it. What's to stop someone else from coming in with a bunch of Dell boxes and cleaning up? Well, the Sun folk were happy to point out that it is one of two companies that have the top to bottom capability to make all parts of the stack, hardware, OS, software, middleware, and just about all other -wares.
Dell assembles cheap hardware, but it has to buy the OS and software. MS doesn't make boxes apart from Xboxes. You can go down the list marking off just about everyone as not having all the pieces, except IBM.
IBM is the new Sun boogyman, it was vilified several times in the conference call. I guess MS is no longer a valid target, something about money talking here, so it is now IBM's turn. I hope the videos turn out as well as the ones about Gates and Ballmer, but somehow I don't think Sam Palmisano is all that great a target for caricature. [Er, we'll see about that. Ed.]
The topic at hand though is the stack, and all parts of it, and IBM is the other company that can do it. Sun was quick to point out that it goes counter to its entire business model. IBM is about selling services and customization for a hefty profit. Sun wants to make everything commodity for, umm, a hefty profit. From the slideshow level overview, I think I prefer the Sun way.
Sun tried to say that even IBM does not have all the pieces, saying that AIX does not run on commodity x86, and there are other pieces missing. I think that Sun is not giving PPC its due, and anyone who thinks that IBM could not make a rock solid Linux distro in a weekend is borderline delusional. Same for any other part of the stack that is currently open source.
That would mean IBM has something that Sun doesn't, a cheap CPU, and Sun has something that IBM does not, the grid control software. Of the two, I would think Sun has it better being able buy commodity chips, there is no commodity large scale grid control software.
Sun did a great job of describing what the grid is and what it does. The whole grid concept has great potential if you have the right type of problem to run on it. If they manage to get a toehold with the big problems, I can see the software evolving to more pedestrian uses. At the very least, it is going to be fun to watch.
Then came the bad, FUD, and lots of it. Much of it was confined to the tail end Q&A sessions, but there was a lot sprinkled liberally in the call. It started out being annoying, and then became borderline sad, and I hope Sun will can it, soon. Fat chance, but hey, I have to ask.
Sun started out sprinkling things in about it having the only indemnified OS out there. There were several instances of indemnity and indemnified OS out there, an obvious crack at Linux and IBM's use of it. It is truly sad that Sun needs to stoop to such pitiful FUD to sell its products.
Sun might be able to say this because the conference call was aimed at business analysts and CxOs rather than thinking individuals, but that does not wash with me. I think it was deliberate and pathetic. It definitely alienates the Linux community, and if that is its goal, it won't be hard to achieve. I don't think it would be a win for Sun though, whether it admits it or not. It needs the support of the community, and antagonism is not a way to make friends. Neither are remarks that make journalists queasy.
Then came the hottest debate raging in open source circles, is Sun serious, or is it just playing semantic games? The CDDL is real, as are many of Sun's contributions to open source. IBM's contributions are real also. Coincidently, both also line up with their respective companies business plans.
IBM is contributing to Linux via GPL'd software and truly open patents. Sun has almost contributed to the community, but there is great debate as to whether it opened it to the general public, or whether it opened it up to the press and people who want to work exclusively on their code.
There are enough flowery press releases and quotes to choke a horse, but when asked the tough question about its intentions, it did the best mustelid impression I have ever seen. If you don't know what I am talking about, read this and this and this for starters, and follow the links for more. To me, the question is, did Sun really open things up, or is it just doing the fashionable PR thing?
That question was unequivocally answered in the Q&A sessions at the end. In my eyes, IBM did the right thing, regardless of its motives. Sun, when asked if it was doing the right thing, dodged the question. When pressed about how open its patents were, it responded with the point that IBM opened a patent on tamper proof screws, intoning that its patents were more applicable.
To me, borderline applicable is much better than closed with a dog and pony show. The clumsy dodging of the Sun execs told me that they Sun is not really opening the patents, or the code.
It's not that Sun is doing the same old thing it always does - that should come as a shock to no one, but the simple fact that it can't admit it. If you are going to sit on one side of the fence, that is fine by us, but don't play such transparent games. Basically, grow up.
For all those waiting for Sun to stop dithering and do the right thing, move on, it won't. It is going to milk the process and it's trailing press for all it is worth. I was hopeful that it would have a change of heart, but the call convinced me otherwise.
The saddest part of this is that until today, I was convinced that Sun was crawling its way out of the pit it found itself in after the .com bust. It re-worked the company, came up with a compelling new strategy, and was on the verge of co-opting one of the hottest trends in software.
I saw the light at the end of the tunnel, but now I question it. I am openly calling on Sun to backpedal here and do the right thing. No weasel words, no hidden clauses, just open things up for real, or shut up and go back to the old ways. I can live with either version, and I think both would be better for Sun.
Sun would be much better served by working honestly with open sourcers.
This quarter's conference call can be summed up in this fashion. First is that Sun has a really neat technology which could potentially change a lot of things. Second it has all the tact in its presentation of a drunken football player on spring break trying to grope a cheerleader. For me, that ruined an otherwise great presentation. µ
Sign up for INQbot – a weekly roundup of the best from the INQ