(moving)
On Mon, Jun 9, 2014, at 04:11 PM, Julian Hyde wrote: Ted is correct that it’s non-trivial to build an adapter to a new data source. I wouldn’t say “very significant” — you can get quite a useful adapter by just pushing down table scans, and doing all other relational operations (filter, aggregate, join etc.) in the client. If you want further optimizations, you need to push down more operations, and that takes thought. But I believe that Phoenix does work reasonably well. After all, it presents a JDBC interface, and Optiq’s JDBC adapter is fairly mature. Gabriel, Bruno, Steven or Karel: Can you give more details about your integration between Optiq and Phoenix? Hey everybody, we can definitely confirm that this is possible. On the entity-side, we have Phoenix, on the customer-level aggregates side, there's Optiq - both accessing data which ultimately resides in HBase. And you can access the Phoenix data using the Optiq JDBC driver transparently. Steven.
