While reading the Helsinki Declaration posts, I got to thinking about a short document on Oracle Performance for Developers I wrote a while back for a US based fund company.
I'd like to share one paragraph:
"An efficient application reads the fewest blocks from memory (logical reads) to perform the work it needs. It has a transaction size (redo size) that contains all the work that must be committed or rolled back together that comprises a business transaction. It does not do any unnecessary commits or rollbacks. It also, parses a statement once, and reuses (executes) it over and over again."
One of the take aways from the Helsinki Declaration posts is "spreading the work unnecessarily across many tiers makes the servers and developers do more work" .
Why do we do this?
