BerkeleyDB performance tuning using perl API

tbusch has asked for the wisdom of the Perl Monks concerning the following question:

Hi Monks,

I'm using the CPAN module BerkeleyDB to share several 1 GB hash tables between processes. While the out-of-box behaviour of those hash tables is fairly quick, I was still wondering if anyone knew of some parameters to improve performance. Do you know what to look out for ? And if yes what parameters do I need to change and how ? I'm sure there a lot of options that are not documented in the CPAN module.

Many thanks in advance for any advice.

Comment on BerkeleyDB performance tuning using perl API

Replies are listed 'Best First'.
Re: BerkeleyDB performance tuning using perl API by samtregar (Abbot) on Jul 28, 2006 at 15:52 UTC
The most important one I know about is CacheSize. A recent change to Geo::PostalCode started setting this to the size of the DB, effectively caching the entire file. The speedup was impressive. Also, if you're using the tied interface you can get a nice bump by switching to OO - tie has a significant overhead in Perl. -sam	[reply]
Re^2: BerkeleyDB performance tuning using perl API by tbusch (Sexton) on Jul 28, 2006 at 16:18 UTC
I had a look at Geo::PostalCode. Unfortunately they use the old Berkeley DB API called DB_File rather than BerkeleyDB. Do you know how to set the CacheSize with the new API ? It's not mentionned in the doc.	[reply]
Re^3: BerkeleyDB performance tuning using perl API by Limbic~Region (Chancellor) on Jul 28, 2006 at 16:39 UTC
tbusch, It's not mentionned in the doc. What docs are you reading? From the synopsis in TFM, `[ -CacheSize => $number, ]` Cheers - L~R	[reply] [d/l]