On Thu, Feb 24, 2011 at 1:34 AM, Michael Kogan
<[email protected]> wrote:

> Understood.  But what topology do you envision to implement the logical 
> streams. Presumably, subscribers and publishers would connect zmq sockets 
> after discovering it with the coordination service/broker.
> Are you thinking of a symmetric solution, where the physical zmq backing 
> behind every stream exists on every machine or a sharded one?  Could 
> publishers on any machine publish to any stream?

Let me give you an example of how this works, as I see it. Imagine we
have a source of earthquake data. The source publishes that as a feed,
"ED". An app wants to subscribe to "ED", so it looks that up, gets an
endpoint, and connects to it. We now have the publisher talking
directly to the subscriber. That app filters the data to produce
earthquake data for South East Asia only, so publishes a feed
"ED-SEA". Another app takes that "ED-SEA" and generates statistics as
"ED-SEA-STAT".

Streams are therefore ways to address publishers (be they original
sources, or filters). I'm still thinking about the best way to handle
multicast, where many nodes can write to the same stream.

This is still rather vague, sorry, it'll be clearer when we have some
running code.

-Pieter
_______________________________________________
zeromq-dev mailing list
[email protected]
http://lists.zeromq.org/mailman/listinfo/zeromq-dev

Reply via email to