So the commercial product is faster? Does it have all the bug fixes that the 
open source prodcut has - I onlt ask I 2.10.10 has fixes in it relevant to our 
code base. (so I have been told)

From: [EMAIL PROTECTED]: [EMAIL PROTECTED]: Mon, 14 Jul 2008 09:31:27 
-0700Subject: Re: [Sequoia] Where could I look to improve performance
Hi Adam, Generally speaking inserts run quite a bit slower than against a 
native database, so if you have a work-load that is pure inserts this sounds 
about right.  We have gotten up to 800 updates per second using our commercial 
product; Sequoia runs at about half that speed.  One of the main differences is 
that the commercial product has faster group communications with some 
specialized performance enhancements.  You can improve performance in the 
following ways: 1.) Eliminate unnecessary SQL calls.  For example, in Hibernate 
you can generate keys using the native UUID generator rather than using the 
database to create keys.   Another trick is to optimize out unnecessary SQL 
like UPDATE statements that don’t normally update any rows.  2.) Don’t use 
transactions.  Transaction commits cause an extra round of messages in GC.  
Using autocommit is therefore faster but obviously not everyone’s cup of tea. 
If these don’t work, you may want to consider uni/cluster, which is the 
commercial implementation of Sequoia.  Check it out at 
http://www.continuent.com. Cheers, RobertOn 7/14/08 9:20 AM, "Adam Purkiss" 
<[EMAIL PROTECTED]> wrote:
I am trying to do a comparison of system performance between our old non 
clustered MYSQL DB and our two controller each with its own DB setup. Both 
setups have a JBoss server running a lot of inserts against it. I am seeing a 
big degrade in performance with the sequoia instance and wondering if people 
out there have ideas on where to look to improve performance? Without sequoia I 
get a 44 second turn around on the test I am running, with Sequoia I get a 360 
second (6 minute) turn around. Now the test may not be fair, I am running a lot 
of inserts and the DBs are replicas of each other but I was wondering if this 
is expected behaviour and where I might look in my configuration to make tweaks 
to improve things.

-- Robert Hodges, CTO, Continuent, Inc.Email:  [EMAIL PROTECTED]:  
+1-510-501-3728  Skype:  hodgesrm
_________________________________________________________________
Find hidden words, unscramble celebrity names, or try the ultimate crossword 
puzzle with Live Search Games. Play now!
http://g.msn.ca/ca55/212
_______________________________________________
Sequoia mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/sequoia

Reply via email to