Accessing Kylin through JDBC driver

Question

I have a Kylin 2.0 cube built from a Hive database (which I access using the Impala JDBC driver). I have added a data source for it using their (Kylin) JDBC driver. The queries are failing. I realize that this is probably out of scope for you, but I would like to ask some questions to help me decide what to do next.

• If I specify "Default Dialect" in my data source, am I to assume that this is a generic JDBC interface? In that case, are there any restrictions on the JDBC driver types that I can use?
• Kylin is a strange system. You query the underlying table which was used to build the cube using aggregation function (count, sum, etc.) and the driver converts these to call to HBase where the data was pre-aggregated. I am seeing Knowage errors that look like cache errors, database not specified, unable to drop table errors (caching I assume). I have tried various ways to get around this, like inserting USE statements, using AS new_table_name, specifying explicit persist tables, etc.

Any ideas? Please note that I really like your product and I think we are trying to push it beyond its current data source boundary, so please do not take my questions as any form of criticism.

Answer 1

It looks as if it is possible to use Mondrian to access Kylin for pure OLAP (i.e. MDX) access. Mondrian handles MDX queries and translates them into JDBC calls that are sent to Kylin. I will be testing this soon. If I make it work I will post information for how to create pivot tables using Kylin. My only issue right now is that the Mondrian MDX jar file was built using Kylin 1.6, but I am using Kylin 2.0, so I may need to do a rebuild.