Hi,

I read an article which recommended to use dataframes instead of rdd
primitives. Now I read about the differences over using DStreams and
Structured Streaming and structured streaming adds a lot of improvements
like checkpointing, windowing, sessioning, fault tolerance etc.

What I am interested to know is, if I have to start a new project is there
any reason to prefer structured streaming over Dstreams?

One argument is that the engine is abstracted with structured streaming and
one can change the micro-batching engine to something like the continuous
processing engine.

Apart from that is there any special reason? Would there be a point in time
when the DStreams and RDD would become obsolete?

Reply via email to