There's more than one way to do things | |
PerlMonks |
Re: Up for Critiqueby Anonymous Monk |
on Mar 23, 2002 at 03:12 UTC ( [id://153712]=note: print w/replies, xml ) | Need Help?? |
One wild suggestion ;-); Don't underestimate the speed of tied DB hashes. As an entry example: $hash{"$nodename"}=$data, $hash{"$nodename".'description'}=$description, $hash{"$nodename".'date'}=$date (if you don't need to split upon retreival), $hash{"$nodename".'date'.'month'}=$datemonth (saves a split upon retreival) etc. By implementing a pseudo java-like namespace scheme you can make some very fast data entry and retreival scheme. Not tested on this ammount of data but certainly tests faster on around 10000 records with many entries (based on / down your figures). There are as ever downsides such as having to write your own retreival routines and then making them interoperable or file locking problems. I believe newer versions of the BSD-DB have transaction support (if you comp dept supports them - they are not free). However you can use any_DB if multiple platform coverage is essential. Plus all of the above (and possibly what comes below too). HLA-B?? rocks!
In Section
Seekers of Perl Wisdom
|
|