See, here’s the thing. I actually like Joel of Joel on Software. He writes well, he’s clearly thought a lot about the process of development and how it should work and the things that can go wrong while doing it, and he’s good at explaining that stuff. We’re using his company’s bugtracking product at work (yeah, even in the admin group we end up developing software, which means bug tracking, and all the bagage that comes with it), and I’m looking forward to seeing what he has the Project Aardvark interns churning out this summer.
That makes it all the more frustrating when he opens his yap to talk about sysadmin-type stuff and a bunch of utter crap falls out (scroll down to the “Network Load Balancing Works” section; they’re called “permalinks”, Joel, look into them). I mean, it’s nice that he’s figured out that you can have two web servers powering your site, and that one benefit of that is that you can upgrade one at a time, and then your web site stays up during the upgrade. (Nice in a “I would have thought the concept of ‘redundancy’ would have come up at Microsoft, but, hey, maybe this explains some things” sort of way, I mean.) But then he goes and says something stupid about the Windows software-based load balancer they’re using prefering to send the same users to the same part of the cluster “so stateful web applications still work even if the state is maintained on one computer.” That was the point at which the respect meter started its rapid leftward motion — a trend that was only intensified when he told me “If you have a hardware load balancer and that needs to be updated or rebooted or if it fails, you’re off the air.” At that point, I realized that the reason TheChild was staring at me wide-eyed was because I’d just yelled “that’s why you have two of them, just like the rest of your network stack, you #$#^@% idiot!” at my laptop.
Seriously, Joel, stick to the software. Leave the enterprise admin stuff to the enterprise admins.