That's right, the requirement for different input source has been asked many times. Kylin 1.x using Hive as protocol and it has been extended to support "SQL on Hadoop" as protocol in Kylin 2.
I think we could leverage the advantages from SparkSQL and Apache Drill (or others), they both offer abstract layer for underline data sources. With the plug-able input source architecture, Kylin2 just need to depends(or include) one library and leave the RDBMs even other sources to them. @Edward, how do you think? Maybe we could consult Drill and Spark community for such idea. Thanks. Luke Best Regards! --------------------- Luke Han On Thu, Feb 4, 2016 at 12:23 AM, KylinPOC <santoshskulkarn...@gmail.com> wrote: > As Luke said: > /But read from RDBMs is valid to extend input source rather than Hive > today, not only RDBMs also SparkSQL, Impala, Drill and other SQL on Hadoop. > / > > This can be a great addition. Many organizations now have Hadoop and other > databases as a data lake/data fabric. > > -- > View this message in context: > http://apache-kylin.74782.x6.nabble.com/Discuss-KYLIN-1351-Kylin-to-support-RDBMS-as-data-source-tp3564p3596.html > Sent from the Apache Kylin mailing list archive at Nabble.com. >
