A lot of hardware RAID systems (whether controller based or SANs) liketo present a multi-disk RAID array to the host operating system as asingle device. While attractive, this can lead to hard to diagnoseperformance issues under load (well, a random IO load).

The basic problem is that you get the best performance by keeping allof your disks busy. But when you aggregate multiple disks into what theoperating system sees as one disk, the operating system is going toschedule IO for one disk

Classic Mistakes Enumerated is an exerpt from the book Rapid Development by Steve McConnell; itruns through 36 familiar classic development mistakes that people makeover and over again. Brooks's Law makes an appearance, of course.

(From Bill de hÓra .)

I found out about Linux su 's ' -p ' option from comments on a a previous entry . The option is also known as -m and --preserve-environment ; the manpage describes it tersely as 'donot reset environment variables', but the info document tells the fullstory:

Do not change the environment variables `HOME', `USER', `LOGNAME',or `SHELL'. Run the shell given in the environment variable`SHELL' instead of the shell from USER'

This is perhaps an obvious observation, but I've recently consciouslyrealized that persistent SCGI and FastCGI daemons area form of caching of regular CGI programs; rather than caching data,they're caching code and often configurations. Thinking about them thisway has made several things clearer to me.

As caches, they have all of the usual concerns with detecting anddealing with cache invalidation, complicated by the fact that fewenvironments have really good support for loading new versions of yourprogram's code on the fly.