> Considering http is a stateless protcol How can you use the
> transaction monitor if the middleware does not understand
> transactions? Is there any advantages? Any comments / web sites would
Take something like an online purchase (cdnow.com). When you enter details
and hit purchase button, lots of things happen behind the scenes, take as
example just three:
[ transaction start ]
1. CDNOW update their stock tables.
2. Financial transaction is made.
3. Flag saying "send this crap to customer x" is updated.
[..]
100. Nothing has gone wrong. Inform user that purchase is success.
[ transaction complete ]
Say the thrid step fails, maybe a sequence has been exchausted or a
constraint violated, or any number of things.
With transaction support, you can rollback and exit with an error (maybe
notifying a dba, whatever). Without.. well, painful..
Mysql kicks bootie for raw speed[1], but the trade off (compared to
postgres) is functionality. That said, I wouldn't consider deploying
either in production.
[1] my p100 with 20M ram - mysql can log every single packet in 56k
connection without hiccups, postgres couldn't even come close.
-- adm
--
SLUG - Sydney Linux Users Group Mailing List - http://www.slug.org.au
To unsubscribe send email to [EMAIL PROTECTED] with
unsubscribe in the text
