[zeromq-dev] Comment on ZRE spec flaw
Michael Haberler
mail17 at mah.priv.at
Sun Feb 24 09:28:31 CET 2013
Am 21.02.2013 um 10:31 schrieb Bjorn Reese:
> On 2013-02-20 10:29, Michael Haberler wrote:
>
>> - a joining node A sends its own node ID, PLUS a list of all node ID's it has already heard (which may be empty at startup)
>
> This got me thinking...
>
> In the following I am not trying to solve a consistency problem, but
> rather to suggest an optimization opportunity.
>
> According to RFC 20, "Each ZRE node SHALL broadcast, at regular
> intervals, a UDP beacon that identifies itself to any listening nodes
> on the network."
>
> If the beacon not only contains information about the sending node, but
> also the peers that the sender is aware of, then those peers do not
> need to broadcast themselves, thus saving network traffic. The list
> should not included all known peers, but only as many as will fit into
> a single UDP package.
>
> Information about peers would have to include their IP addresses.
>
> Peer disappearance is handled via heartbeating.
OSPF uses the neighbor lists for two purposes:
- to build a tree of router adjacencies
- to assure communication between neighbors is actually two-way
I think the first purpose isnt terribly relevant for ZRE
the second possibly could be useful to fight protocol autism ;)
- Michael
>
> PS: What is the rationale for creating the Zyre discovery protocol
> rather than reusing one of the existing discovery protocols (ZeroConf,
> SSDP, SLP, etc.) ?
> _______________________________________________
> zeromq-dev mailing list
> zeromq-dev at lists.zeromq.org
> http://lists.zeromq.org/mailman/listinfo/zeromq-dev
More information about the zeromq-dev
mailing list