posted on Jun, 17 2009 @ 02:22 PM
It's that time again. After about a year of steady duty, our database server has recently experienced what can only be described as the upper-limits
of it's capabilities. We knew this about four months ago, and have since been stashing away enough savings to spring for a brand new server, and an
industrial strength database expert to configure it.
Our latest database server has done well, lasting longer and handling more traffic than any before it. Peaking at an astonishing 318 queries per
second at one point, it brought us through Blossom Goodchild, the 2008 elections, Swine Flue, and very nearly double the average daily traffic since
it was deployed.
And we're not just getting a new server, we're beginning the transition to a cluster of database servers. Initially all activity will be handled by
a lightning-fast speed demon master server, with a single slave server acting as a read-only failover should the master go offline. For those into
such things, the new server has 16 cores, 8 gigs of RAM, and more importantly, SSD's (multiple solid state drives) in a RAID-10 configuration. Our
slave server(s) will be less powerful, but still us SSD's for storage. This is the bleeding-edge folks.
We're currently scheduled to perform the transition Friday morning, June 19th (because our expert doesn't work weekends), and if all goes well, most
of you probably won't notice. However, if you encounter ATS in a read-only state, or even offline on Friday morning, please don't be alarmed and
check back in 15 minutes.
Onward...
[edit on 17-6-2009 by SkepticOverlord]