[zeromq-dev] HWM in multi-speed connection scenarios

Andrew Hume andrew at research.att.com
Sun Sep 2 23:52:45 CEST 2012


the HWM applies at each end of teh connection.
so it is sort of global from the (each) publisher, and local at each client.

again, 0MQ is NOT a substitute for a properly fair job scheduler.
(although under many conditions, it works fine for this purpose.)


On Sep 2, 2012, at 1:00 PM, Parham Negahdar wrote:

> I have a question in regards to the HWM. Lets say you have a publisher publishing data at 1Gbit/s and three subscribers, one who can recieve at 100Mbit/s, one at 500Mbits and one at the full 1Gbit/s. And you set the publisher's  HWM to 10 messages. Now is the HWM global or on a per client basis? So say there is a sudden burst. Now the queue will already be filled up from the messages that the slow clients haven't received but the fast clients have. Is the queue smart enough to know which connections have already received the data? Will the queue for the slower clients get replaced by the burst data needed by the fast one? Is there any documentaiton on how the queue is handled and how smart it is. Essentially how does ZeroMQ behave when publishing data really fast, to clients of varied connection (not processing) speed? 
> 
> 
> 
> Already read this document: http://zguide.zeromq.org/lua:chapter5 
> 
> 
> Thanks!
> 
> 
> _______________________________________________
> zeromq-dev mailing list
> zeromq-dev at lists.zeromq.org
> http://lists.zeromq.org/mailman/listinfo/zeromq-dev


------------------
Andrew Hume  (best -> Telework) +1 623-551-2845
andrew at research.att.com  (Work) +1 973-236-2014
AT&T Labs - Research; member of USENIX and LOPSA




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.zeromq.org/pipermail/zeromq-dev/attachments/20120902/bc76d4fb/attachment.html>


More information about the zeromq-dev mailing list