Minimum hardware capacity for 35k e-mail scans/day

Jason Ede J.Ede at birchenallhowden.co.uk
Mon Nov 19 09:26:10 GMT 2007


I'm guessing those benchmark tests are for 1 process trying to get the lock... Also if its file locking thats the issue then if you've 1 or 2 MailScanner children then won't the berkeleyDB be faster and the advantages of SQL bayes will increase as the number of children increase (and sql means other machines can also access DB)

Jason

From: mailscanner-bounces at lists.mailscanner.info [mailscanner-bounces at lists.mailscanner.info] On Behalf Of Steve Freegard [steve.freegard at fsl.com]
Sent: 16 November 2007 17:19
To: MailScanner discussion
Subject: Re: Minimum hardware capacity for 35k e-mail scans/day

Julian Field wrote:
> Greg Matthews wrote:

>> UxBoD wrote:
>>> also what type of filesystem is the Bayes on, if it is in a file, is
>>> journaling switched on etc etc ... to be honest since switching it
>>> into MySQL I have been very pleased with the performance.

>> I find it amazing that mysql is faster than BerkeleyDB can anyone
>> explain this? BerkeleyDB should knock the spots of a full featured
>> RDBMS shouldnt it?

> Only 1 process at a time can hold the BerkeleyDB open and write to it.
> Lots of processes can hold the RDBMS open at once. It's all down to file
> locking.

See http://wiki.apache.org/spamassassin/BayesBenchmarkResults

Cheers,
Steve.
--
MailScanner mailing list
mailscanner at lists.mailscanner.info
http://lists.mailscanner.info/mailman/listinfo/mailscanner

Before posting, read http://wiki.mailscanner.info/posting

Support MailScanner development - buy the book off the website!


More information about the MailScanner mailing list