Comments on Tom Pinckney: Use it or lose it

Sounds like your QA scripts will likely find any d...

2010-01-10T19:57:28.758-05:00

Sounds like your QA scripts will likely find any data corruption issues. My question was more out of curiosity, because any time I've done a restore I've never been able to do anything but spot check the data. I have, at times, thought about calculating checksums of key data before the restore for checking after, which sounds like what the utility you referenced does. Good stuff!

This is great advice for a software development/IT...

2010-01-10T17:29:14.940-05:00

This is great advice for a software development/IT environment. Non-IT related, I’ve heard some pretty funny stories about power generators starting up when there was a power outage and they either failed to turn on or, worse, they set on fire!

@Louis -- The restore gets tested when the develop...

2010-01-10T17:21:01.203-05:00

@Louis -- The restore gets tested when the developers sit down in the morning to use their dev environment or when the QA automation scripts run over night. This doesn't catch everything, but so far whenever there is a problem it's usually pretty catastrophic and shows up immediately.

What are your thoughts on how we could do this better? We've thought about ussing Maatkit's checksum utility but never gotten around to it.

Things rot. That is so very true. Code rots. Infra...

2010-01-10T17:15:51.066-05:00

Things rot. That is so very true. Code rots. Infrastructure rots. Integrations rot. IT is one of the most direct, real world examples of entropy I can think of.

Great post. Quick question: on your DB restores, how do verify data integrity? Do you assume that if the DB restored OK then the data is OK, or do you perform 2nd level validation of the restored data.