February 15, 2017 at 9:21 am
Looking for real world experience here. I have a oltp system that tracks medication administrations in long term card. It has several tables involved in this transaction (between 30-40 million rows), total size only about 60 gig. Administrations done at assigned pill times so very busy at specified times. So looking at possibility of moving tables to memory. From realworld experience which are the best candidates for this and what types of tables should be avoided?
Thanks
February 15, 2017 at 3:10 pm
Can't help with the experience, but with 30 to 40 million rows, the first thing is going to be ensuring that you have enough RAM to avoid starving out other SQL processes, so that you don't end up going to disk because there wasn't enough RAM hanging around to keep that table in memory. Not sure how SQL Server manages RAM in that case, but what could happen is a RACE condition as processes end up constantly swapping RAM to DISK and back because you run out of RAM resources and go CPU crazy just trying to swap it all in and out. After all, with just a 1 KB row size, you're looking at 30 to 40 GB of RAM to hold that table in memory.
Steve (aka sgmunson) 🙂 🙂 🙂
Rent Servers for Income (picks and shovels strategy)
Viewing 2 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply