- A "t2.micro" serving the site. This is managed using ECS (the EC2 Container System) and a Docker container built from the source code on GitHub. We're probably going to automate building and deploying using AWS CodePipeline at some point, so we'll just be able to push to a GitHub branch and update the site's software.
- A "db.t2.micro" hosting the database. This is a dedicated database VM managed and backed up by AWS.
The database seems to be ticking along pretty nicely without too much load, at least for the moment:
We're not using very much CPU—only about 5 to 10% of what we've been allocated. But I'm a little concerned by the IOPS. We currently have 100 IOPS, and it looks like we're already using a third of that. Again, there's a burst mode available, up to about 3000 IOPS, but once we use it up, we go back to 100.
In both cases, we can upgrade the underlying servers if we want to go faster (or downgrade to save money). It would take literally a couple of hours. But faster servers cost more money. You can play around with the various server sizes (look in the "Linux on Demand" column for the price) and database server sizes (look in the "MySQL on Demand" column for the price).
So, how's forum performance looking right now? Is everything reasonably fast and snappy? Do you ever see sudden, massive slowdowns, as if we exhausted our CPU credits or IOPS?