All geek questions in one place

MDX support for Hive (Hadoop)

Is there support for multidimensional expressions (MDX) for Hadoop Hive?

+6
source share
3 answers

It is possible to connect an OLAP solution with Hadoop data. In icCube you can create your own data sources (check documentation ), you will need a Java interface (for example, JDBC).

This solution transfers data to the OLAP server. Transferring processing to Hadoop is another question, and as far as I know, nobody does. Possible simultaneous aggregation of facts. Another step is to measure in nodes. This is a complex problem (algos are not easy to transform in a parallel version).

+1
source

You can use Mondrian (Pentaho Analysis Services), it connects via JDBC and uses certain dialects for databases. I saw a link to the Hive dialect, but I haven’t tried it myself - it’s best to look for forums.

There is a little learning curve: you need to create a schema that defines cubes in XML, but, fortunately, this is a GUI tool (schema toolkit) that helps.

0
source

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

0
source

Source: https://habr.com/ru/post/888794/

More articles:


All Articles