Re: [dnssd] multicast over wireless links

[Robert Cragie <[e-mail]>]

+1. Of course 802.15.4-based mesh networks have to cope with multicast 
traffic but its use should be carefully considered.

Robert

On 24/07/2014 7:43 PM, Ted Lemon wrote:
> On Jul 24, 2014, at 2:33 PM, Robert Cragie <[e-mail]> wrote:
>> But as Alf said earlier, there are other wireless technologies which are not well suited to IP multicast traffic. 802.15.4-based mesh networks are a very good example and there is certainly a use case for DNS-SD in these type of networks.
> Broadcast is a good idea when the information being distributed is needed by all receivers, and when receivers' connections are persistent.   But newcomers to the network don't have the information until the next broadcast, or have incomplete information until the next broadcast, or have to trigger a new broadcast.
>
> In practice, in my experience, what this means is that multicast winds up producing much more traffic than would a non-multicast solution.   So that's why I suggest that it's better to rely heavily on multicast only in environments where multicast really is cheaper than unicast.
>
> (Of course, clearly DNSSD will use multicast to some extent; the question is how much.)
>
> _______________________________________________
> dnssd mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/dnssd
>