Monday, September 22, 2008

Browser process models

Everyone's praising IE8 for it's new one-process-per-tab model. It's got many advantages over the threaded model used by most browsers, including the fact that a crash of one tab can't take other tabs or the browser down. What most people don't seem to get is that the multiple-process model isn't new with IE8, and in fact the threaded model was one adopted only over the strenuous objections of everybody else. You see, threads exist for one reason and one reason only: on VMS and Windows (which was designed in part by the guy responsible for designing VMS), creating a process is expensive. You need threads in those OSes because you can't create a lot of processes quickly. But Unix had cheap process creation from day one. You needed another thread in Unix, you forked off another process. Threads weren't needed. But everybody in the Windows community kept braying about needing threads and why didn't Unix have them, oblivious to the fact that they already had what threads did in the fork() call. So Unix finally, reluctantly, adopted threads with all their pitfalls. And programmers used them heavily when processes would've been more appropriate. Until the pitfalls finally became too much to live with, when they went from just driving programmers nuts to causing problems for the very users who demanded them in the first place. So now we're back to where Unix was 25 years ago, forking a new process when you need a new thread of execution.

Thursday, September 18, 2008

Cooling a data center without AC

Intel has done a test, cooling a high-load data center model without using air conditioning, just ambient outside air. They did see a higher failure rate on the equipment, but not nearly as much higher as was expected. And the portion that used a simple, cheap cooling system without all the climate control of a true data-center AC unit had a lower server failure rate than full-on data-center-class AC yielded. My feeling is that it's not the cold air that's critical to data-center health, the servers will be entirely happy with 80-90F cooling air. It's mostly the dust and to a lesser degree the order-of-magnitude changes in humidity that cause most of the failures. Scrub the dust from the air (even cheap AC units do this, and simple filter systems can do it too) and keep the humidity in the 5-35% range (no need to control it to +/- 2%) and you'll give the servers 95% of what they want to be happy. And you can do that for a fraction the cost of a full climate-control system that controls everything to within 1%.

Tuesday, September 16, 2008

Need to update site

I really need to start updating Silverglass (the Web site) again. It used to have a lot of pages about the nitty-gritty of configuring RedHat Linux. I've long since switched to Debian, and the configuration tools have gotten a lot better since the days when getting e-mail address masquerading working involved manually editing, but there's still things I'd like to document. Setting up the zone files for a DNS server, for example, and configuring a DNS "blackhole" list (a set of DNS domains you want to make not exist anymore, eg. DoubleClick's domains, so that Web advertisements and such just vanish). And some things haven't changed, setting up a firewall for example, and details of Debian's startup scripts and how to play nice with them when writing your own is still useful information. And of course there's rants I want to write but haven't, the originals would be on my blogs of course but copies can go on the Web site. Then there's code projects like the LJBackup program I can finish up and drop on the site.

I just need to get time and energy together and actually do it.

Monday, September 15, 2008

Hubble Space Telescope discovers unidentified object

The HST has spotted an unidentified object. (Here's a more serious article.) It's not similar to any type of object previously seen, and it's spectrum doesn't match that of any known object or type of object. There wasn't anything in that part of space before it blipped in, and there's nothing obvious there now that it's faded. So what is it? Nobody's got any good ideas yet.

"The most exciting phrase in science, the one most likely to herald a new discovery, is rarely "Eureka!". More often it's "That's funny. It's not supposed to do that, is it?"."

Wednesday, September 10, 2008

Large Hadron Collider

They've fired up the LHC today for the first tests. A lot of people have been making noises about the risk of it creating a black hole that'll swallow Earth. I'm sorry, the world won't be ending today.
  1. Yes, the LHC will be colliding particles with higher energies than anything humans have been able to manage to date. But the Universe isn't human, and has been colliding particles with energies orders of magnitude higher than the LHC's capable of for billions of years. Several such collisions happen here on Earth each year as high-energy gamma rays impact the atmosphere. Given the length of time and the number of events per year, if those collisions would create a black hole that'd last any length of time we'd've seen evidence of it happening before now.
  2. Any black hole the LHC might create will have only the mass of the particles involved in the collision. That's only going to be a couple of protons worth. Such low-mass black holes emit a large amount of Hawking radiation relative to their mass. And that emitted radiation comes from their mass. Low-mass black holes simply evaporate very quickly (within fractions of a second) after forming. So even if a black hole does get created, it'll disappear again probably before we even know it existed.
  3. Even if the black hole sticks around, it won't pull in enough to be a problem. Remember, black holes don't have any greater gravitational pull than anything else of the same mass, their only special property is how steep their gravity well is. The danger radius of such a low-mass black hole is going to be a fraction the size of a subatomic particle. It's going to have to hit another subatomic particle almost dead-center to suck in any additional mass. At this scale "solid' matter's 99.999% empty, so it's going to take that black hole a very long time (as in millions of years) to accumulate enough mass to begin to start pulling in matter on a macroscopic scale.
  4. Ignoring the above, the black hole will have the same velocity vector and momentum as the particles that created it. That velocity'll be well above the Earth's escape velocity, and tangent to it's surface. Any black hole will simply continue in a straight line away from Earth, never to return.
  5. And even ignoring the previous points, this is the test phase. They've only turned on one beam to calibrate and align things. With only one beam, there aren't going to be any particles colliding. So even if the LHC could create black holes, it won't be creating them today.
So, I'm sorry, but the crowbar-wielding Gordon Freeman won't be getting any screen time because of the LHC.