To Seshu, currently we are refactoring Job Engine to support both MR and Spark, so we don't have any patch yet(it is on my own fork). JIRA to track: https://issues.apache.org/jira/browse/KYLIN-744 And to Maria, yes kylin will be able to support data source other than hive like SparkSql, Spark-Streaming hopefully in the next few months. Will keep you updated when there is some progress.
Gaspare Maria <[email protected]>于2015年7月15日周三 下午10:35写道： > Hi, > > very good. > > Does it mean we can use spark and (when needed) spark-streaming to load > cubes in kylin without use Hive ? > > Hi have many use-case where I use spark-streaming to process, aggregate > and load on HBase raw data coming from kafka. > > In same cases I need also MOLAP, so could be perfect to use kylin and my > ETL based on kafka + spark-streaming. > > Let as know. > > Thanks. > > Regards, > > -- gas > > > On 07/15/2015 03:09 PM, Adunuthula, Seshu wrote: > > Cool!. Do you have a JIRA, have you submitted a patch? > > > > > > On 7/15/15, 5:55 AM, "Luke Han" <[email protected]> wrote: > > > >> Awesome job, go go Spark... > >> > >> > >> Best Regards! > >> --------------------- > >> > >> Luke Han > >> > >> On Wed, Jul 15, 2015 at 6:10 PM, 周千昊 <[email protected]> wrote: > >> > >>> after some test, we are now able to > >>> *write hfile through spark and then bulk load into htable* > >>> We are using the spark api 'saveAsNewAPIHadoopFile' to achieve that. > >>> Does anyone has any suggestion on that or have any better solution? > >>> > > > >
