Clunk Werclick wrote: > On Thu, 2009-07-23 at 13:50 +1000, Barney Desmond wrote: >> You need to ask yourself if this is a real problem, or something >> you're just imagining. Mysql generally works fine, 50,000 messages a >> day at 12 queries each, equates to several queries per second. This is >> an "easy" load. > That is a comfort to know. My main concern was this hammering was not > optimal, but it is welcome to make as many queries as it likes if it > does not crash the database server. Perhaps Postgresql would be a bit > more manly ? but slower ?
You'll probably not note a difference. I guess MySQL will allow you to connnect() faster if using a local socket. However you should always use proxy_read_maps - so connect()-times are not so relevant. I gave a quick look at the server statistics of our MySQL instance providing Postix and Amavis config (not used as Amavis storage etc, its only purpose is providing "configuration"): DB uptime 250 days with an average of 300 queries per second (our reports are showing peeks of slightly more than 6 million delivery attempts a day). We are using multiple servers, but that's mostly as of disaster recovery and failover reasons - you could handle similar traffic also on a single host (using recent server hardware). A certain percentage of queries could of course be avoided if Postfix where optimized for DB usage. As we know it isn't - this design choice however keeps it flexible and simple. Best regards, Thomas Gelf