We are using IPC::MM and it works great. We use it to cache about 5000 strings for our internationlized systems ( EFIGS-J right now, going to 15 languages soon ). Its pretty easy, in our startup handler we have :
my $MM_SIZE = 5000000; my $MM_FILE = 'st_cache_mm_file'; my $st_cache_mm = IPC::MM::mm_create($MM_SIZE, $MM_FILE); my $st_cache_btree = IPC::MM::mm_make_btree_table($st_cache_mm); tie %CACHE::mmcache, 'IPC::MM::BTree', $st_cache_btree; Then we access it via the $CACHE::mmcache variable by language ID/Message ID hashes. The performance is pretty good, a huge huge improvement over direct to database calls. The way we have it configured is that each child calls a function per string. This function checkes the apache wide MM cache, if it doesn't find it then it grabs it from the DB and sticks it there for the next child to come along. We were using 'per child' caching but it was slow, expensive and gave page load speeds with a lot of variance ( .5-1.5 seconds for cache, 10 seconds for uncachced ). It was pretty nasty. We are not really expiring our cache, we just HUP apache to re-initialize it since MM is local to the root apache process. Your mileage may vary, ours is pretty good. John On Sat, 05 Oct 2002 15:40:05 -0300 Cristóvão Dalla Costa <[EMAIL PROTECTED]> wrote: >Does anyone have experience with Apache::SharedMem? I'd >like to use it to store an in-perl cache of a few >thousand database items, in order to decrease load, but I >noticed that it's version 0.09 and not updated in a year, >so I became a little suspicious. Any comments? > >Thanks. >
