Amazon has released its Relational Database service today. It has the full capability of MySQL and the service lives in the Cloud, as its other services like S3 or SQS, etc do.
Also similar to other services, it is charged by usage.
Why it is a great news? A couple of things came to my mind when I heard it.
RDS can be your single DB server. If you are a web developer who manages a small/medium site on shared hosting server you won’t have the luxury to build a standalone DB server and you’ll likely end up with Apache and MySQL running on the same box, and you know it is a pain in the butt when moving a site from one host to another (transferring files, re-establish DB, and probably making code changes to point to the new DB). With RDS, you can simply create your DB instance in the cloud and skip moving DB when you need to move your hosting service. This may sound only benefit your hosting company since one less use is competing for the resource, but you definitely get the advantage of flexibility and probably some increment on reliability.
However, moving DB to cloud will increase network latency, and as a developer I have my share of concern on the Cloud-based services. It’s out there and I have very little control of. We have had a few times of S3 going offline in the past and it’s not fun for those services or sites whose architectural diagram live inside the cloud service itself. So for a large site owner with a “mission critical” database, I would recommend using RDS to cut down the cost of setting up DB mirrors. For instance, we often using a second MySQL DB instance to sync with the master; and RDS can replace the slave in this case. It’ll be useful for failover, backup and real time restore, and a little bit of downtime is probably okay. The biggest benefit is that we can save a lot by eliminating the cost of server, software management and system admin work.
It’s certainly a bad news for service like FathomDB, which is a cloud relational DB service built based on Amazon’s EC2. This probably makes a good argument against making a long term commitment on building services which rely on platforms that in other people’s control. When a nice Facebook app get popular, who’s to stop Facebook to build the similar one for their own as the market has been proven? Same goes to iPhone apps. Make some quick bucks and move on is the way to go.
Any other ideas of using RDS? Let me know. Cloud is certainly making things cheaper for us web builders.
