On 4/26/06, Daniel Kinzler daniel@brightbyte.de wrote:
cool!
this will allow MySQL to run without influence from CPU-consuming user processes.
Hm... how much CPU eating user processes do we see currently? Does this split really make sense?
I would propose to split a different way: have a "research" or "analysis" server (few but heavy queries) separate from the "tools" server (web service, many relatively fast queries). I believe this would help more to make the server(s) more useful.
I'd propose something slightly different..
We run two servers:
1. Back end, live replicated databases. And mature 'production' tools which require greater reliability. 2. Front end, most tools that are not so sensitive to toolserver load, all tool development, and additional analysis databases which are not live updated.. but which may contain extra indexes, and other fields to help some tools work.
Thoughts?