SFX fixes for scalability

As an FYI … we’re doing a couple of things to make SFX whole again as well as get the rest of the OSL-hosted Mozilla infrastructure on solid footing.

This morning we repurposed one of the hosts serving up www.mozilla.org to be a full-time database server. This is a very robust piece of hardware and this machine will serve as the primary database server for SFX. Moving forward, we will be migrating the databases for other services hosted here at the OSL (I mention the OSL only because we wanted to have a DB server local to the services hosted here). More on this in a bit.

Now we have seperate caching, application and database servers for SFX and the site appears to be much snappier and scalable. We’ll be watching the new setup over the next couple of days to make sure everything is working appropriately.

We have several other services (download.mozilla.org, UMO, etc) hosted here that we need infrastructure for and to that end we will be purchasing equipment to shore up our existing applications as well as scale it even further. This will entail a redesign of how we’re delivering our services to folks; most of this should be entirely transparent to the end users. The purchase of this equipment will also allow us to move machines around to needed services as they see increased load/usage/growth.

Once this new equipment is here, we will be repurposing the machine that is currently serving up UMO and turning it into a backup database server doing replication with the existing DB server.

Sorry for all of the recent SFX outages; this is on the radar now and will hopefully be much more stable moving forward.