Hi Nunzio, all!
I cannot give specific hints, not being a MySQL tuning expert, but I repeat my general question: Nunzio Daveri schrieb: > Hello Gurus :-) I was running a simple load generator against our 16GB Dual > Quad core server and it pretty much came down to it's knees within two hours > of > running tests. The customer DOES NOT WANT to change any code, they just want > to > throw hardware at it since it took them a year to create all of the code. *Which* *component* exactly "came down"? In order for others to give helpful hints, you need to tell whether it's CPU or disk, the cache sizes you have, and (preferably) the hit rates. Also, do "vmstat" while the server is loaded - in the worst case, your caches are larger than your RAM will hold, and you get paging added to cache reload. > It > is > a 140GB database with 21GB of indexs all using InnoDB - currently doing 70% > reads and 30% writes. That is at least some info. Personally, I think 30 % writes is a relatively high rate, and I'm not sure replication will help. > > My question is what is the best way of distributing the load without changing > any of the php / perl code that their web server uses? This is what I am > thinking but need someone to tell me it is a good idea or bad please? Before you decide to distribute the load, you need to check whether the limiting component in the current machine can have its load reduced or its capacity increased. Expanding a component is sure to be cheaper than adding another similar box. Also, are you sure there is no other activity on that machine which could be taken off? > > [[...]] Jörg -- Joerg Bruehe, MySQL Build Team, joerg.bru...@oracle.com ORACLE Deutschland B.V. & Co. KG, Komturstrasse 18a, D-12099 Berlin Geschaeftsfuehrer: Juergen Kunz, Marcel v.d. Molen, Alexander v.d. Ven Amtsgericht Muenchen: HRA 95603 -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/mysql?unsub=arch...@jab.org
