[zeromq-dev] Proposal regarding PGM and ZMQ

Martin Sustrik sustrik at 250bpm.com
Sat May 15 07:51:16 CEST 2010


Steven McCoy wrote:
> On 15 May 2010 01:04, Pieter Hintjens <ph at imatix.com 
> <mailto:ph at imatix.com>> wrote:
> 
>     On Fri, May 14, 2010 at 5:40 PM, Martin Sustrik <sustrik at 250bpm.com
>     <mailto:sustrik at 250bpm.com>> wrote:
> 
>      > 1. RDATA cannot take precedence to ODATA
>      > 2. ODATA cannot take precedence to RDATA
> 
>     Perhaps this is more sophisticated than we can do, but it would seem
>     that the ideal design is to allow RDATA to take precedence over ODATA
>     but only up to a certain %, say 10%.  So if there are no errors, RDATA
>     consumes 0% and ODATA gets 100%, but if the network or clients are
>     unreliable, the RDATA consumes more, but only up to a safe limit. The
>     RDATA limit could probably be hardcoded for almost all cases.
> 
> 
> RDATA as a percentage is fine, I can use 
> "source_transmit_rdata_max_percentage" as per SmartPGM,

Ah, is that already a part of OpenPGM?

> ODATA over RDATA is good for real time data, especially with pro-active 
> FEC which is transmitted as ODATA.
> 
> RDATA over ODATA is effectively congestion control.
> 
> Both PGMCC and pro-active FEC are half implemented in v3, there is no 
> implementation of RDATA over ODATA.

Martin



More information about the zeromq-dev mailing list