Yep, this is exactly the case. And unfortunately DB optimization is WAYYYYY above my pay grade…

Right on. With more RAM I could increase the size of several very important caches that can save a lot of database requests, which are usually the most CPU-intensive.

The main issue is that the VM we’re running on has CPU burst capacity, but recently we’ve been bursting it almost all the time. This will soon lead to the host throttling us down to the baseline performance that I’m paying for, and when that happens a small spike in requests will flatline the whole system and basically bring the site to its knees. So I’m now looking at upgrading to a different service with dedicated CPU cores and a lot more RAM. But meanwhile I had to take the GT thread down to avoid using up all of our burst allowance.

I guess we’ve been doing pretty good thusfar with the fairly limited server resources we have. BLF gets well over 1 million hits per month (usually somewhere around 30k/day), and that’s just counting users that don’t run ad blockers (my only source of traffic stats is Google Adsense).