Quote:
Originally Posted by photon
I think it was more that there was actually a networking problem somewhere, at least in this case, but I understand what you are saying about the longer term stuff.
Forums are extremely database intensive.. so it used to be there was almost no choices, but more recently places like DigitalOcean are using SSDs for their VPSs/VMs which help a lot.
There is an upgrade plan in place, and it's almost completed, I had just thought the performance issues had been mostly mitigated so I had put it further down the priority list (my monitoring was telling me that it was getting 3 second load times).
But I guess the monitoring isn't telling the whole story.
|
Yeah SSD's really aren't necessary if the VM's are hosted on decent SAN's. Any mid tier EMC/Hitachi pool would have more than enough IOPS to absolutely dominate any about 50 database backends. It's not like it's SAP
As you said, definitely don't think it's a DB performance issue, the connection just fizzes out. Some nerd is probably upset at one of the other sites hosted on the same connection and chucks his piss poor little botnet at it. That's where having scalable/redundant network infrastructure is a winner.
I'd just find a little script to monitor the sites latency, should tell the tale.