Github user industrial-sloth commented on the pull request:
https://github.com/apache/spark/pull/3858#issuecomment-113631381
Sure thing @tdas. First a caveat: I haven't been keeping up with the spark
community since ~March 2015, so the issues I originally hit might no longer
exist w/ more recent spark releases.
As of December 2014 we were exploring streaming options for real time
analysis in Thunder (https://github.com/thunder-project). Thunder is pyspark
based; at that time our pyspark dstream options, as I recall, were basically
either file-based (watch a directory for new files) or to integrate with Kafka.
Specifically there was no option to listen to a ZeroMQ stream or to many of the
other dstream types available in the scala API. We wanted to be pushing a
high-bandwidth stream of microscope images over to pyspark for further
analysis. ZeroMQ seemed ideal; Kafka seemed like too much and file-based seemed
to necessitate an additional unnecessary disk IO. So I put together a ZMQ
solution for pyspark streaming and threw it out there in this PR.
Again, haven't been keeping up, not sure whether this is still a concern w/
current releases of pyspark. I agree this is potentially an unusual use case -
our workaround at the time was to go to the file-based dstream implementation,
which was functional but perhaps not optimal.
Any further comment on this @freeman-lab or @andrewosh?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
