October 9, 2007 · 7:30 PM
Sean Chittenden presents:
Mr. DBA, tear down your IO!
Mr and Mrs developer, for the liberalization and prosperity of
your database and application, come understand your IO! Mr
Developer, open your eyes! Mr DBA, come tear down your IO!!" Of
course I'm paraphrasing, but what does it mean to your RDBMS, OS
and infrastructure when you start mashing out transactions? How
expensive is an SQL operation in terms of IO and how much IO is
done for different datasets? Come understand a sample of IO
profiles for PostgreSQL for small (millions of rows) and large
(hundreds to billions of rows) data sets accessed via internal,
NFS, and SAN. What's the difference in IO between a SELECT,
indexed SELECTs, clustered indexes, internal storage, and a
storage array with a boatload of cache in a lab environment?
RSVP ASAP to get fed :)
Thanks very much to Arcscale for hosting.