I am evaluating a choice between Oracl开发者_运维问答e OLAP and Pentaho Mondrian.
At the same time some people say that using a column DB could simply make the use of OLAP's redundant as they are much faster.
has anyone got any experience on the same. Will it help if ou OLAP sits on a column db ?
OLAP is not about the storage only; it is about the MDX language as well. MDX is much more powerful than SQL to handle multi-dimensional (with hierarchical dimensions) queries.
You can have a look to www.icCube.com - it's pretty fast - I mean no problem to aggregate let's say 30 millions of facts over 20 dimensions in sub-seconds (without any pre-aggregation computation).
Columnar databases are truly designed for analytics. Beye Network published a series of blogs about columnar databases/when to use them. One of them can be found here: http://www.b-eye-network.com/blogs/mcknight/archives/2010/06/the_value_of_pe.php.
In full disclosure, I am the community manager for Infobright, an open-source Columnar database company. Pentaho and Infobright work very well together; in fact, there's a virtual machine with both Pentaho/Infobright installed. That VM can be downloaded on www.infobright.org.
Cheers and good luck,
Jeff
精彩评论