I'm about to design a web application in Perl, to run on an Apache server (Linux), which will require a large, possibly hundred megabyte, in-memory data structure. I would like to initialize the data structure once from a database or file, then allow updates to affect the structure, as well as the persistent datastore, so that all future HTTP processes can access and update the most recent data.

I've tried using mod_perl for this type of persistent, in memory data, but any data structures that are initially shared get copied once per process and become unshared as soon as the data is changed by a process.

Do any other Monks have experience with using shared Perl data structures among multiple Perl (Apache) processes? What Perl, Apache, and/or Linux options are available handling this?