Hi Walid,
Granted it’s a little hard to understand without a picture, but the sentence is 
correct.  Consider this:  While a tuple is being processed by one of the 
parallelized bolts, it remains un-acked.  So if there are ‘N’ parallel bolt 
instances, and you want them to all stay busy, then the system must allow at 
least ‘N’ unacked tuples to be in existence without complaining.  Actually you 
should allow ‘N + a few’, so that you have a bit of queue to truly keep them 
busy.  If you only allow ‘Q’ unacked tuples, where Q < N, then only Q parallel 
bolt instances can ever be busy at once; the Q+1’th un-acked tuple coming into 
the topology causes the limit to be exceeded.

Hope this helps,
--Matt

From: Walid Aljoby <[email protected]>
Reply-To: "[email protected]" <[email protected]>, Walid Aljoby 
<[email protected]>
Date: Friday, November 25, 2016 at 7:04 AM
To: User <[email protected]>
Subject: Number of unacked tuples

Hi Everyone,

Could anyone please explain this sentence by an example? I got confused with 
word "lower", I think it should be "higher".

If the number of possible unacked tuples is lower than the total parallelism 
you’ve set
for your topology, then it could be a bottleneck.

Thank you,
--
Regards
WA

Reply via email to