Wow, I was thinking about something like SimpleDB two days ago.
SimpleDB looks awesome. It actually looks a lot like CouchDB. Maybe that's what they are using? They are charging per "CPU cycle", so they could have CouchDB servers running in EC2 and are measuring the resource usage.
Going back to my design.. I was thinking of doing something like a json+memcached+S3 storage engine. The API would store and retrieve data using json. On lookup, it would check a memcache first. If it didn't find the object in the memcache, it would pull the object from S3 and stick it in the memcache.
Of course, my system doesn't support searching, and you'd have to build another layer for indexing/referencing data. I'll be happy to use SimpleDB.
Now Amazon needs to implement a load balancing and static IP's and I'll never have to buy a server again.
No comments:
Post a Comment