Performance did not get substantially worse yet since last reboot on January 29th.
My guess one of these things could help:
----We ran UPDATE STATISTICS 2-3 times since last reboot
----Moved Transaction Log file of one of busy databases to a proper drive (before it was sitting together with data files)
Maybe I was wrong when said performance gradually gets worse.
I'm afraid it might be pretty sudden.
Like 2-3 days of slower than usual and then big slowdown.
We don't see a big Disk latency now. Like we saw last time before reboot.
Overall it stays on the same level it was right after reboot.
Noticed the thing about performance being suddenly worse instead of gradually...
We had an issue a few months back with one of our servers, where the VM host was writing a journal file - it basically tracks all of the changes that have occurred on the disk since the last snapshot. Anyway, once the file would get to a certain size, things would very abruptly slow to a crawl on the SQL instance. Our hardware team explained that it was because on the storage it was having to scan through this huge file for basically every transaction. Sorry if that's vague but I'm not a hardware storage guy. In any event, they disabled the journal and we haven't had the problem since.
That's a good point, as well. I've seen some might strange things along that line. We had a tried and true job on an AS 400 that normally only took an hour to run. One morning, it changed to 6 hours. No code changes were made. Nothing was rebooted. No large amounts of data were imported, It was crazy... after a week, they gave up and called IBM for some help. To make a longer story shorter, the battery on the cache for the box had finally given up the ghost. They replaced it without even taking the machine down and performance instantly returned to normal.
So, yeah... I agree. It could be something like you say.