One of our more stable clients had some serious “mystery” production problems. To paraphrase, their app that was running “OK” starting running “sucky”. I knew that nothing had changed from our end and so being the intrepid super-sleuths that we are, we determined that the problems were related to I/O.
This first installment will deal with the biggest contributor to bad application design — the dreaded bad query. I’ve actually had busy systems crash because a developer unleashed a query that performed a full tablescan too many.