Hi, It's a good question, and in short, I think Phoenix and Spark overall serve orthogonal use cases.
If all of your needs are met with the Phoenix JDBC interface, you likely have no need for Spark at all. However, if you have data in HBase/Phoenix and would like to leverage Spark functionality, such as MLLib or GraphX, then the phoenix-spark integration provides a way to do so. The phoenix-spark documentation has a brief example of applying PageRank using GraphX to a Phoenix table [1] Best, Josh [1] https://phoenix.apache.org/phoenix_spark.html On Wed, Nov 16, 2016 at 10:20 PM, Christopher Tarnas < [email protected]> wrote: > It is not an either or, you can use both - hence the plugin. Phoenix is > great at OLTP type workloads and Spark is better at OLAP and machine > learning. > > -chris > > > On Nov 16, 2016, at 6:56 PM, Cheyenne Forbes < > [email protected]> wrote: > > > > so why would I choose Phoenix over Spark? >
