I haven’t had a chance to do more posts or a wrap up of SC06. I will do this soon. I want to briefly point out this article. And offer a mea culpa.
Briefly, I had a discussion with Patrick of the Microsoft team about Microsoft’s goals and vision. You know, if you just remove the CEO’s occasional statements about his competitor being a virus, a cancer, and so on, the vision isn’t bad, and is something that we can work with. Unfortunately there are lots of missing elements. If they can give us the tools we want (Patrick knows which ones I was talking about, and which ones I would like to continue to talk about), we can do some pretty awesome things. More about the vision in a moment.
My major complaints about the Microsoft approach to HPC comes from 2 major locations. First is what I call the “replacement hypothesis”. That is, that Microsoft marketing is positioning itself in such a way as to suggest that now that it is thinking about HPC, HPC is mainstream, and the great unwashed masses will come running away from the pathetic stone tools they have been made to endure until now, so they can run Excel across a cluster. Because thats what every one wants, to run excel across a cluster. Well, not quite. Critical to their message is that Linux is hard: hard to use, hard to install, hard to administer. It is unfortunate for their message that all of these assumptions are incorrect. Moreover as part of this “replacement method”, they talk about how MCSEs will happily configure clusters of the future. Sure, 2-4 machine clusters. I can believe that. 8 way and higher? I don’t think so, but if you want to, what will result is actually more business for us. Part of our business is re-engineering and fixing the mistakes of others. You should see some of the more interesting MCSE-generated fundamental cluster design mistakes we have fixed in the last 90 days. This doesn’t mean those machines can’t run windows, just that if you are going to design something to scale, it helps to have a rather deep understanding of what “to scale” means in the context of the application and how it impacts the design. This is not what an MCSE has. This is what we have.
Second point that I made to Patrick was, after he suggested that I get the HPC bits from the HPC team (reasonable request), I noted that more than occasionally the critical execs pretty much undermine any credibility of the HPC team about interoperability and playing well together, by opening their mouths. See here.
It seems fairly obvious to me now that the execs seem to be plotting not how they are going to interoperate with a competitor that their customer wants them to interoperate with, but how they are going to leverage the legal system to do what they cannot do in the market. This would be sad. Business is business, and you have to do what you have to do. Rather than continuing to use the soon-to-be-bankrupt SCO as their proxy in the fight, MSFT ought to just have at it. Start up the legal shenanigans. Lets see a listing of exactly what IP Linux has “misappropriated”. Microsoft does have a patent portfolio, and quite unlike SCO, actually owns something. I would be surprised if IBM or SUn have to pay Microsoft for each copy of AIX or Solaris shipped. If there were Microsoft owned IP in there, thats what would need to happen.
I don’t have a problem with Microsoft promoting windows, that is, to a degree, part of their focus. I (and the market numbers that I am aware of from IDC and elsewhere) strongly suggest that windows is not growing at the expense of Linux as Mr. Ballmer suggests, but rather the other way around. Linux has been and continues to decimate Unix. Disagree if you want, show me the massive upswing in Solaris 10 installs. I haven’t seen it (apart from Sun shipping it on every unit, most of which on x86_64 are wiped then to install Linux).
We are customer focused. Our customers are growing the Linux support, and windows support. They are taking down *all* of their unix support. They are in cost reduction mode: TCO cost, and again, unlike the Microsoft funded studies, end user data reported in many places (google for it) suggests strongly that the TCO is not in Microsoft’s favor. This in part explains why Linux server growth is rather significantly outpacing the growth of the server market in general. This has been pointed out here. Windows server growth is about the same as the market as a whole. This doesn’t sound like the scenario Mr. Ballmer implies.
Again, I am somewhat saddened to hear this. I expect the lawsuits to start flying soon. Microsoft has effectively infinite resources with which to defend itself and execute offensive tactical legal maneuvers. With what Mr. Ballmer said about IP and the wording of the SuSE agreement, yeah, it all clicks into place. Seems rather obvious now.
Ok, so here is the challenge. Lets see precisely which IP Linux allegedly purloined from Microsoft, or is using without a license. Pointers to patents at www.uspto.gov would be most appreciated.
When SCO came out with its lawsuit, we discussed this with our customers: it was obviously FUD and had very little likelihood of having even a grain of truth to it. We were curious as to who put SCO up to such an idiotic course. Who was driving SCO ahead of it as an auxillary? Three years in, we know that answer. Follow the money, connect the dots, we know that this was and is a well funded effort to disrupt Linux uptake, primarily by two organizations. One whose product offerings are being decimated by Linux, another who is feeling threatened by Linux. Only one of our customers bought into the line, and has since come back to tell us that he understands now that this was simply FUD.
Now we see the sabre rattling yet again. And again, the accusations of IP mis-appropriation have started. And again, an indemnification of sorts has been offered.
Rather than compete in the marketplace, they plan to compete in the courts.
They appear to be planning to use their overall market dominance to suppress and inhibit the growth of a competitor, through the court systems, and not the marketplace where they are not being successful against this competitor. Something about this familiar. We have seen this before. Using one’s market leverage to exclude others from the market is proscribed behavior for a company that has an effective monopoly. Then again, I am not a lawyer, so ask one if you want to know the correct meanings.
The SuSE deal is not a good deal. I see where Microsoft is heading. This is a shame, as the windows HPC initiative could make for wonderful interoperability with the massive and growing Linux HPC market, and nascent (and effectively invisible) windows HPC market.
I was wrong about the SuSE deal. My apologies. We will use SuSE where our customers require it, and we will move away from it everywhere else. Just like windows, it will be applied only where absolutely needed, and no other better/cheaper/stabler/more-secure alternative exists.