I started speaking a year ago and yesterday at SQL Saturday Exeter was my 6th session. Looking back on it, I tried to cram way too much into the allocated fifty minutes. I rammed it in, had to deliver the material too fast and got flustered and struggled with some of the questions due to the time-stress. This annoys me, as I feel I didn’t do the best that I could. The session in future will be less than half as big, with plenty of room for questions. Learn, speak, learn, repeat.
Thinking back, I think I confused Hugo Kornelis the most, whilst trying to explain how Always On worked with in-memory data. Hugo, I apologise for this. It has annoyed me so much that I felt compelled to write this quick post to clear things up.
Always On AGs and In-Memory OLTP
- Log records are transferred to the secondary replicas and replayed for all types of tables: disk-based data to disk, in-memory data to memory (so on failover there is no need for a reload of the in-memory data as it is already there and ready to rock)
Oh, and talking about bleeding edge stuff your testing but haven’t finished yet, such as controlled failovers for in-memory maintenance? Get them sorted and tested first, talk about them second. I learn….
This was where I possibly confused Hugo and every other attendee by talking about “memory-mapping”. Fool. Why did I plump for that phrase? This will NOT be used in future.
I’m going to have a week off thinking of sessions now but when I get back to it they’re going to be smaller and more focused. Less is more. Quality over quantity. I learn, I speak, I learn, I repeat (the speaking, minus the crap bits and with more good bits).