Game On

Recursive blogging category: I can now let you know about a new feature at Server Pipeline -- my "blog" is now actually a blog. This will let me bring you more pointers, comments, suggestions and the like on a quick and regular basis, in addition to the longer column-type pieces you've been seeing there. And it's set up for comments now, too, so I'll be wanting to hear what you think about things, even if you didn't agree with something I said. Maybe especially if you didn't agree. Who likes echo chambers? So bring it on, people -- we'll all benefit as a result. (And don't expect it to be 100% servers, either; if I see something fun or interesting that seems good to pass along, I'll do that too. After all, all work and no play makes all of us dull boys and girls, no?)

It looks like a few things happened while I was away. My only interaction with the Zotob worm was reading about it in a newspaper, that ancient, virus-free form of communication. And you can bet I liked it that way. But to me, the real news du jour continues to be AMD's and Intel's efforts at one-upsmanship in the server processor arena. My colleague Jennifer Bosavage was, of course, right on about the implications of Intel's dual-core moves last week (and many thanks to Jen for the great pinch-hitting), but this week adds on the speed and performance wars. Moore's Law is still not failing us in the server space: Intel is looking to get performance out of its new Xeons that can more than triple the loads it can handle in a given time, and AMD is pushing the limits of the Opteron just as hard.

Unlike their legal battles or other skirmishes, though, this is news you can use, especially if you're running a large data center. It's kind of like owning a DVD player that breaks -- why get it fixed when it's cheaper and easier to just get a newer, better model? Similarly, with those kinds of gains in chip performance, administrators are gong to have less incentive to tweak the odds and ends of their network and server setups; when you can get that much more right at the chip level, why bother? Throw in multi-cores, and CIOs and server admins are likely going to be looking at a Golden Age of efficient, widely distributed data center performance. Start saving those capital budget pennies now.