RE: bayes DBM versus SQL

2006-03-02 Thread Gary W. Smith 
DBM is fine for small installations but if you need to scale up then SQL
will allow for a consolidate database across multiple machine.

We use it on a decent size platform (multiple front end relays, multiple
sa boxes and a clustered MySQL instance).  It works well for us.

For bayes training, we use a separate linux workstation and just point
it to the master database.  This gives us the ability to do tasks
without impacting the overall SA boxes.



 -Original Message-
 From: Webmaster [mailto:[EMAIL PROTECTED]
 Sent: Wednesday, March 01, 2006 11:59 PM
 To: users@spamassassin.apache.org
 Subject: bayes DBM versus SQL
 
 Those of you you have used both native DBM and new SQL bayesian,
 can you comment on benefits of one versus the other please.
 
 Much appreciated!

Re: bayes DBM versus SQL

2006-03-02 Thread Steven Stern 

Webmaster wrote:
Those of you you have used both native DBM and new SQL bayesian, 
can you comment on benefits of one versus the other please.


Much appreciated!



I  have three MX servers fronting our Exchange box.  The fastest of the 
MX servers is also handling the MySQL server for both bayes and AWL. 
It's surprisingly fast and all three boxes are working from the same set 
of information so the path the mail takes doesn't affect scoring.  Most 
of the spam comes through the non-preferred MX server.


--

  Steve

RE: bayes DBM versus SQL

2006-03-02 Thread Webmaster 
Gary W. Smith wrote

  -Original Message-
  From: Webmaster [mailto:[EMAIL PROTECTED]
  Sent: Wednesday, March 01, 2006 11:59 PM
  To: users@spamassassin.apache.org
  Subject: bayes DBM versus SQL
  
  Those of you you have used both native DBM and new SQL 
 bayesian, can 
  you comment on benefits of one versus the other please.
  
  Much appreciated!
  
 
 DBM is fine for small installations but if you need to scale 
 up then SQL will allow for a consolidate database across 
 multiple machine.
 
 We use it on a decent size platform (multiple front end 
 relays, multiple sa boxes and a clustered MySQL instance).  
 It works well for us.
 
 For bayes training, we use a separate linux workstation and 
 just point it to the master database.  This gives us the 
 ability to do tasks without impacting the overall SA boxes.
 
 

Thanks.  
So SQL is the way to go for cluster enviroment.
Having a separate training box is also a good idea.

RE: bayes DBM versus SQL

2006-03-02 Thread Webmaster 
 

 -Original Message-
 From: Steven Stern [mailto:[EMAIL PROTECTED] 
 Sent: March 2, 2006 5:13 AM
 To: spamass
 Subject: Re: bayes DBM versus SQL
 
 Webmaster wrote:
  Those of you you have used both native DBM and new SQL 
 bayesian, can 
  you comment on benefits of one versus the other please.
  
  Much appreciated!
  
 
 I  have three MX servers fronting our Exchange box.  The 
 fastest of the MX servers is also handling the MySQL server 
 for both bayes and AWL. 
 It's surprisingly fast and all three boxes are working from 
 the same set of information so the path the mail takes 
 doesn't affect scoring.  Most of the spam comes through the 
 non-preferred MX server.
 
 -- 
 
Steve


Thanks for letting me know Steve.  That's what I want to hear. 
I will implement SQL version of Bayes then.

bayes DBM versus SQL

2006-03-01 Thread Webmaster 
Those of you you have used both native DBM and new SQL bayesian, 
can you comment on benefits of one versus the other please.

Much appreciated!