I found on the agenda of the MySQL Conference & Expo 2007 that Mark Atwood presented in his talk A Storage Engine for Amazon S3,how to use the Simple Storage Service to store the data of a MySQL database.
That MySQL allows to exchange the storage engine with plugins is a well-known fact. That’s how the transition from MyISAM to the transaction-capable InnoDB storage engine was made. But to move the storage completely on the internet by using S3 is an unusual, yet interesting idea.
Amazon offers S3 as
a simple web services interface that can be used to store and retrieve any amount of data, at any time, from anywhere on the web.
S3 users don’t have to deal with RAIDs, storage volumes, scaling capacity, doing backup or any other technical details. It’s all taken care of at the Amazon data center.
I’d be interested how well this internet storage system can be used as the storage backend for MySQL — what about latency, bandwidth and other issues? For example, what happens if I do a full table scan over a big table that’s stored on S3?
Maybe there’s some MySQL expert out there that can shed a bit more light on that topic?