Hekaton (aka. In Memory OLTP Tables) is awesome. Things run very fast through the system, pushing you to 98x the speed of normal tables.
But be careful. You’ll need to know well in advance how much transaction log space you are going to need. You see there’s a little gotcha that I ran into recently (thankfully in a lab and not a production server). If the transaction log fills because say the hard drive of the server runs out of drive space, the transactions don’t fail to write, the database goes suspect.
Neat trick hugh.
So you’ll want to make VERY sure that you’ve got PLENTY of disk space for the transaction log and that you are backing that bad boy up often otherwise some really bad things are going to happen.
You’ll need to be even more careful if you are using Availability Groups as the Availability Groups feature queues up the data which needs to be replicated within the transaction log. So if a replica goes offline for a while you’ll need to have enough space on all the servers for the transaction log to grow until someone can get to the Availability Group and take the failed replica out of the configuration.
Denny
