MDX support for Hive (Hadoop)

Is there any support for Multidimensio开发者_如何学Cnal Expressions (MDX) for Hadoop's Hive ?

---

Connecting an OLAP solution with Hadoop's data is possible. In icCube it's possible to create your own data sources (check documentation), you'll need a Java interface (like JDBC).

This solution is bringing the data to the OLAP server. To bring the processing to Hadoop is another question and at my knowledge nobody does it. Aggregating the facts in parallel is possible. Another step is to have the dimensions in the nodes. This is a complicated problem (algos are not easy to transform in a parallel version).

---

You can use Mondrian (Pentaho Analysis Services), it connects via JDBC and uses specific dialects for databases. I've seen reference to a Hive dialect, but have not tried it myself - best to search the forums.

There is a bit of a learning curve: you need to create a schema that defines the cubes in XML, but fortunately there is a GUI tool (schema workbench) that helps.

---

There is Simba MDX provider which claims to convert MDX queries to HiveQL. I have not tried it myself to comment on the features and limitations of this.