Original URL: http://www.theregister.co.uk/2009/07/30/rethink_mysql/

MySQL startup targets SSDs

Stop moving!

By Gavin Clarke

Posted in Applications, 30th July 2009 00:13 GMT

Need a MySQL storage engine tailored for solid-state drives?

Start-up Hexagram 49 has unveiled an engine dubbed RethinkDB. The company says it's optimized for SSDs, claiming it can deliver performance ten times faster than existing databases.

RehinkDB is the work of computer science students Michael Glukhovsky, Slava Akhmechet, and Leif Walsh, who promised their database would simply drop into the popular MySQL. They warned too that RethinkDB is still at the pre-alpha phase.

RethinkDB has listed an early roadmap and invited people to participate here. The team promised more information on the database during the coming weeks.

The work is part of a growing trend to tune applications, particularly databases, to the underlying hardware in PCs and servers to speed performance and cut down on additional hardware. Ingres this week announced the VectorWise Project to put a high-performance storage engine into its open-source database tuned to multi-core, multi-threaded chips.

SSDs have been emerging from Intel and others for use in PCs and servers as an alternative to traditional magnetic hard drives.

The advantage of SSDs is that they take up less physical space, consume less power, and provide potentially faster read and write times because there are fewer moving parts.

They do not require motors and power to spin up and drive the heads, with storage taking place in the processor and in a circuit.

But traditional database engines have not been architecturally geared to the different read-write-and-access techniques that SSDs employ compared to traditional drives.

RethinkDB claims to have solved the problem using a set of "elegant and simple" algorithms that improve performance.

When modifying data, RethinkDB changes only the end of a storage file, meaning potentially faster storage and access because there's less data to physically write. This also means there's no need to lock data because readers and writers can access data concurrently.

The database provides the ability to make schema changes, garbage collection, and backups while the database is live, and there's no need for a log file. The RethinkDB team said their database is the log file, so there's no need to write or manage additional schema. ®