Against the advice of critics everywhere, I saw A.I. last night, and I’m still angry. What a dumb movie!
Here’s just one ridiculous, unbelievable detail that leaves me banging my head against the wall: a genius scientist creates a robot that is in nearly every respect indistinguishable from a real 11-yr old boy, and yet when that robo-boy swallows some food, he short-circuits. Err, can you say design flaw?
This detail isn’t central to the story, making it even more insulting.
If you haven’t seen the movie, take my advice: wait until it comes out on the rental market, and then rent something else instead.
Another day, another exploit of a Microsoft product. Sigh.
Today’s installment brings us the SirCam virus, which emails itself to everyone in your address book (multiple times!) and overwrites critical system files.
This doesn’t concern me particularly, but I want to mention a great anti-virus product that works at the server level. If you run a UNIX mail server, consider installing AMaViS to have all inbound mail scanned for virii prior to delivery.
This is a remarkable product. I’ve seen it in action; today it corralled over a dozen copies of SirCam on a busy corporate server.
Once or twice a year we throw a “pizza and Zinfandel” dinner party, where we eat lots of pizza and drink lots of Zinfandel. Last night’s episode succeeded in both respects. Here, in the style of Harper’s Index, is a recap:
I attended a presentation by two of the lead systems geeks from Google last night. The content dipped below my threshhold of interest when it got into the minutiae of their net-booting protocols, and we nearly throttled one guy who kept asking questions that were way too detailed to be of general interest (especially when it became evident that the guy asking questions had no idea what he was talking about).
Google is hosted on 10,000 Linux servers, most of which are half-depth 1U devices from rackable.com (no link — I thereby spare you the pain you’d feel from rackable’s awful Flash-based website). Each host has 2 Maxtor 40- or 80-gb IDE (!) drives accessed primarily via DMA. They didn’t use SCSI because it’s too expensive, and they can get comparable performance with DMA/IDE. Surprising!
Their web index and cache is replicated across many servers, meaning that a single “database server” has only a tiny fraction of the entire index available within local hardware. This is fascinating, and I guess it’s the heart of their scaling technology — to be able to break down a single index across many servers and query this cluster quickly and effectively.
Many of these architectural details are contained in this article: Google Defies Dotcom Downturn
In a humorous twist, Google presented attendees with Google-branded boxer shorts emblazoned with their appropriate and familiar slogan “I’m Feeling Lucky.”
I am stunned by the speed with which the Code Red worm propagated across the web.
I do not run any Microsoft server products, so my systems are not vulnerable to this particular attack, and yet the worm hit all my sites repeatedly in an attempt to replicate itself.
debris.com was hit from 16 different infected hosts. monauraljerk.org was hit by 26. An unnamed, unpublished, empty website that I use only for testing purposes got hit 18 times.
This is a good reminder to everyone who runs a server to keep up-to-date with vendor patches and bug reports. Subscribe to CERT and BugTraq. That’s the minimum sane level of paranoia; you’d be much better off actually becoming well-versed at server security, or hiring someone who is, to make sure your servers are at least difficult to hack.