in reply to OO Perl & RDBMS Strategy Question

I cache all accessed object information in memory, and whatever is modified is stored back in the database at the end of the CGI call. It also sets some flags in the database that tell other processes if they have to reload their object data into memory. This works better if you have lots of reads and just a few writes. It doesn't work well for very large databases, unless you have a very smart cache control, or lots of memory.