There is a good direction in database land away from the idea that tuning is all seat-of-the pants, smoke-n-mirrors stuff. In Oracle land it was the recognition of Oracle's wait events and the ability to see what Oracle was waiting on and hence discover a probable bottleneck. This replaced the hit ratio and other ratio crowd as the logical and objective nature of the using Oracle waits proved more benefical.
There is still a way to go, as there is plenty of good work done on areas which directly relate to database stuff. Unfortunately for lots of people, math causes some serious FUD (Fear, Uncertainty and Doubt).
I use sage as my RSS feed controller in firefox, whilst the list of new blogs grows, the ability to read content in a single sitting doesn't diminish. I came across an article from one of the couple of math related blogs. This blog talks about using Markov Chains to model resource pools.
I will be having fun tomorrow gathering the necessary information to check the database connection pools which have databases I support. A couple have excess numbers of pools (which waste memory resources) and with better modeling we should be able to recommend some lower numbers.
Have Fun
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment