Mark Semsel's picture

Column-oriented databases have become a hot topic in the Business Intelligence / Data Warehousing arena (see http://en.wikipedia.org/wiki/Column-oriented_DBMS ), and there might be some real interest if there were a Turnkey Column-oriented DBMS appliance.

There are several open-source column DBs available, so one question is which one would be most practical to invest in.  The first ones that come to my mind are Infinidb, InfoBright and LucidDB, but Ingres, a long underappreciated open source DB has a column-oriented project called "Ingres VectorWise".

But is it possible to create a 64-bit Turnkey Linux appliance?  I'd want to think along those lines because column dbs are at their best with massive volumes of data, and they'd benefit from the memory advantages that a 64-bit implementation would have.

Perhaps the easiest way to build this would be to follow the architecture of the MySQL appliance and install one of the MySQL-base column-oriented DBs.

Any thoughts on whether this is worth looking into?

Forum: 
Liraz Siri's picture

Regarding 64-bit support, it's planned and we plan on shifting focus to that after we push out the Lucid based builds.

With regards to other databases, this is definitely worth looking into. The open source database world has more to offer than MySQL and PostgreSQL. There are NoSQL databases such as Cassandra, document oriented databases such as MongoDB and CouchDB, and column-oriented DBs too. We'd like to eventually have TurnKey appliances available for all of them,  and the more help we get from the community the faster it will happen.

Ideally we'd get the community around those other projects involved to help provide guidance and maybe put together a proof of concept TKLPatch that works out the integration details.

Developing new appliances for the best open source projects is something we're personally interested in so eventually even without community help it will happen but as you can imagine we have limited resources and can't do everything at once ourselves.

Add new comment