Two Steps Back
The Register has a post here about another Microsoft Windows security update being pulled because it creates problems greater than it solves. The update cuts all access to the Internet because of an interaction with third-party anti-virus/firewall software.
This is similar to the last update, noted in my April 23rd post (see it here) where I made a brief mention about the problem because it seemed to be affecting one of my PCs. The problem could be resolved only after disabling the ZoneAlarm firewall that I was using. Essentially, the PC came to almost a complete halt while CPU utilization went to 100 percent.
The original 811493 patch I talked about has been reissued here. There is also a Knowledge Base 819634 article here.
I don't know when the latest problem will be fixed, but one must wonder about how much regression testing is done before these updates are released. I realize that there has to be a balance between rushing a fix to a security flaw versus testing the fix to ensure it does not create its own problem. But to modify the old saying; "If you can measure the number of errors you are creating, you are creating too many."
Aloha!
Comments
The main problem with patches, especially MS patches, is that they don't solve a problem.
Typically problems are being solved continuously but in stead of sending out each solution as a patch that are all lumped together. Each solution is individually tested (on the clean target they are reported on) but the combination is probably never tested.
Add to this cocktail the MS tendency to throw modifications and 'enhancements' into the mix and you have a guarantee for disaster.
"If MS says it's error_free they mean you get free errors"
Posted by: sjon | May 29, 2003 03:24 AM