As we always like to tell our clients, Sourcetoad is not a devops company. That isn’t our core competency. However, that doesn’t mean we don’t know about devops and provide devops as ancillary services to some of our clients. We just don’t provide it as a main service.
Our smaller clients don’t even need to worry about devops. We provide simple hosting for their application at a very reasonable rate. It’s also fully managed in the sense that they don’t need to worry about servers, infrastructure, etc. On occasion, some smaller clients will approach us with a question regarding Amazon Web Services. The question usually is as follows: “AWS is free for a year for a small server with the free tier, then only around $10 a month after that. Why can’t I use it for hosting?”
On the face of it, the statement is true. Someone with a fair amount of Linux knowledge could get an AWS instance running a small site for around $10 a month. In fact, we have a test server that operates this way. For a production environment, this would be an extremely bad idea. It’s also against all of AWS’s own recommendations.
AWS is not a normal hosting company like Rackspace, LiquidWeb or Godaddy. AWS is more of a set of tools to allow someone with a lot of knowledge of these tools to build a robust cloud infrastructure.
Let’s start with the example above. If you did setup a t2.micro instance on the free tier, installed Amazon Linux on it, then setup the web server/database/etc., you would have something akin to a small VPS from a normal hosting company. This would be completely unmanaged. You would have to ensure it is operating efficiently and pulling the appropriate security updates. You would have no one to call to ask any questions regarding the software itself. If the instance failed (and they do), and then didn’t come back up, you would also have no one to call. You would have to make a posting on the forum and hope AWS addresses it in a timely fashion. You’d also have to be prepared for them to say, “the instance is lost, recreate it.” You would then have to figure out how to do that while restoring your data. There are no automatic backups. More than likely your EBS drive backing the instance won’t be lost, but you’d need to figure out how to get that up on a new instance. AWS offers phone and e-mail support, but that is a lot more expensive than $10 a month. The lowest level also probably wouldn’t help with the above task.
At Sourcetoad, we haven’t lost any of our instances on AWS, but they have gone down for hours. Rebooting them did nothing. We waited and they eventually came back up. Again, these are all test environments, so it didn’t really matter. On production sites, that is not acceptable.
Stay tuned for part two which describes the proper way to setup a site using AWS’s services.