Several days ago, I enabled change tracking on a db and some of its tables.
A few days ago, I also SET ALLOW_SNAPSHOT_ISOLATION ON
.
(Probably important note: I never SET READ_COMMITTED_SNAPSHOT ON
, and the only procedures/queries that actually used transaction isolation where a few manual runs by myself, which I saw them run to completion within a logical time frame).
A few days after that, we got reports of decreased performance. I am aware that both of these things impact performance in a minor(?) way, but i am not sure how to measure this yet. So, for the sake of being on the safe side, I SET ALLOW_SNAPSHOT_ISOLATION OFF
, and also disabled change tracking on every table first and on the whole db in the end.
The performance problems remain. So my questions are:
- Can change tracking and/or ALLOW_SNAPSHOT_ISOLATION cause performance drop in a server/database even AFTER they are disabled? If so, how, and how can I check and fix the problem?
I run a few queries from this site to check the state of tempdb but did not find any interestingly large table.