Atleast and atmost once guarantees is the key, its hard to achieve that in your setup.
On Tue, Apr 18, 2017 at 1:30 AM Mohit Gupta <success.mohit.gu...@gmail.com> wrote: > Hi, > > Sorry for sounding naive, but I am trying to understand the use cases of > storm. The rationale given in storm documentation is mainly the pain in > deploying/managing queues & workers and partitioning the work between the > workers. (http://storm.apache.org/releases/current/Rationale.html ) > > But now that we have reliable and established queues like Kafka & others, > it looks simpler to me to have multiple queues and multiple workers > depending upon load processing messages from each queue. > > Eg. Word Count Push tweets to a queue ( q1 ) -> multiple workers > processing this queue and pushing words to another queue ( q2 ) -> a worker > generating word count. > > Can you please share an example of a production topology (with slight > comparison) where it would be complex to handle to traditional queue/worker > model but easier with storm? > > Thanks & Regards > Mohit >
