Database Mirroring Freak Show – SQL Quiz from Chris Shaw
Chris Shaw ( Blog | Twitter ) started another great web chainpost. He tagged Brent Ozar, Brent tagged Thomas LaRock and Tom tagged me. This particular quiz is a two-part question and here are my responses. Excuse my lack of wit and charm. It’s late, my caffeine supply is running low and my cat is eyeballing me in a most peculiar fashion.
Do you feel that you have a reliable SAN solution? If so, what’s the secret?
It’s hard for me to answer this one given that just a few weeks ago our data center (SAN included) came crashing down hard. Now given that the problem was a faulty generator test and not the SAN itself that was the problem I can’t really place blame there. Honestly I can’t say anything really bad about our SAN. Tons of disk space, tons of cache, it does what its supposed to do. We’re also in the middle of finding a replacement for a SAN administrator so what SAN solutions we have in place now could radically change in the coming months. As far as secrets go I’d say you just need to make sure you have an open line of communication with your SAN administrator. They have no idea what’s going on in your world and you have no idea what’s going in theirs. Clear communication of needs need to be there as a SAN admin worth his salt is going to know what they need to do on their end to make sure you get the best performance on your end (i.e. proper RAID levels for your LUNs depending on needs, I/O throughput, etc.).
Explain Database Mirroring in layman’s terms
Everyone else seems to be giving off-the-wall answers to this so I’ll give it a go as well. Not going to lie, took me a good portion of my drive to Melbourne last week to finally come up with an example.
Think of clustering as a conjoined twin. You’re talking to the same body. Both heads can hear the conversation but ultimately the two are stuck together because they have to share the same base trunk. Now if you were to punch one twin in the face and knock him out you’d still be able to talk to the other head but you’re still lugging around that base. Now, imagine a set of regular twins (non-conjoined). This is your database mirroring in that you have two separate entities. You knock one out but the other is still chugging along just fine. Only difference being that the location of the second one doesn’t really matter because he doesn’t have a shared trunk to deal with. I’m sorry if that explanation sucks, if you’re bored check out the whitepaper written by someone who doesn’t have a fascination for using genetic defects to compare feature sets. Ok time for me to tag a couple of
unwilling victims fellow bloggers: