Hi Christian, I've just commented on the zeppelin ticket. Thanks!
On Wed, Jul 29, 2015 at 1:37 PM, Christian Tzolov <[email protected]> wrote: > This discussion started in another post but i've decided to move it in a > dedicated thread. > > The Geode Interpter JIRA ticket: > https://issues.apache.org/jira/browse/ZEPPELIN-189 > > > From: @William Markito (my comments are inline) > > @Christian: Thanks for creating that. What do you mean by non-OQL ? > > By non-OQL i meant GFSH shell-proxy (or alike) interpreters. Not sure how > feasible or useful this would be? > > > So talking specifically about Zeppelin and Geode, technically it's > already possible today using the Geode spark connector. I'm going to blog > about it soon, but in a nutshell the connector allows you to execute OQL > within a spark/sql context andregister them as RDDs: > > For example: > > %spark > > val deviceLocations = sqlContext.gemfireOQL("select d.x, d.y, d.deviceId > > from > /DeviceLocations d") deviceLocations.registerTempTable("deviceLocations") > > Then allows you to execute spark SQL such as in this image: > http://img42.com/En7hO > > This looks great. In this way you can share state between Spark and Geode > which makes this approach very practical. As far as i can see > Zeppelin's paragraph does not share state unless one usages and external > store. > > > That said, OQL is a more direct access to Geode and both access would > have different use cases... > > Indeed the OQL interpreter can be useful for fast checking and exploring > the Geode data and then one can use OQL queries in the Spark with Geode > paragraphs. > > Cheers, > Christian > > -- > Christian Tzolov <http://www.linkedin.com/in/tzolov> | Solution > Architect, EMEA Practice Team | Pivotal <http://pivotal.io/> > [email protected]|+31610285517 > -- William Markito Oliveira Enterprise Architect -- For questions about Apache Geode, please write to *[email protected] <[email protected]>*
