0

I have two nifi streaming source in flink, I need to perform joins on this two sources. Which will be the better solution to do that? Is it joins api provided by DataStreams or table api(https://ci.apache.org/projects/flink/flink-docs-release-1.5/dev/table/streaming.html#streaming-concepts).?

Also how to maintain the state of the streams and joins? Is there any in build solution provided by flink.

Thanks, Rilwan

Rilwan
  • 88
  • 11

1 Answers1

1

It very much depends on what computations do you want to apply on your joined data. If you can express your computations with StreamSQL (or table api). I would advice to use it, as it abstracts all the burden of low level details of the implementation like e.g. state maintaining.

If the Table API is not enough then switch to the DataStream API.

Dawid Wysakowicz
  • 3,402
  • 17
  • 33
  • Before asking a question please go through documentation first. Sections regarding state in flink describes it quite well: https://ci.apache.org/projects/flink/flink-docs-release-1.5/dev/stream/state/ – Dawid Wysakowicz May 30 '18 at 07:08