Re: I-D Action: draft-ietf-mboned-ieee802-mcast-problems-05.txt
"Holland, Jake" <[email protected]> Sat, 11 May 2019 00:52:27 +0000
| Newsgroups | gmane.ietf.mboned |
|---|---|
| Message-ID | <[email protected]> |
AFAIK it depends a lot on which particular router. Some of them are based on OpenWRT or LEDE (not always obviously), and from the patchworks message on the initial upstream submission, it looks like those have had default enabled support for mc2uc on AP interfaces since the builds from at least about early 2016 or so: https://patchwork.kernel.org/patch/9494055/ I don't have a platform handy this week to check how well that seems to hold up on the ones I can check easily, but if I remember I'll try to take a look and send an update next time I get a chance. Of course, this is different from "most customer routers", so I'm not contradicting Mikael here, just mentioning a point that might be relevant to the question. And I agree that a very common case is likely to be high loss with a high use of airtime, at least for the routers where you can get them to transmit the packets at all (in my anecdotal experience, it seems common to block the traffic by default except when it's sourced inside the nat). -Jake On 2019-04-29, 13:22, "Jeffrey (Zhaohui) Zhang" <[email protected]> wrote: I was hoping to hear that, even w/o the unicast conversion, since the router does not have to use a low speed to reach many receivers (of different signaling levels), it would not be problem. Guess I was too naïve 😊 Thanks. Jeffrey Juniper Internal > -----Original Message----- > From: Mikael Abrahamsson <[email protected]> > Sent: Monday, April 29, 2019 4:17 PM > To: Jeffrey (Zhaohui) Zhang <[email protected]> > Cc: [email protected]; [email protected] > Subject: Re: [MBONED] I-D Action: draft-ietf-mboned-ieee802-mcast- > problems-05.txt > > On Mon, 29 Apr 2019, Jeffrey (Zhaohui) Zhang wrote: > > > I am also curious, in a home environment, if only one receiver joins a > > particular multicast flow (e.g. one viewer for a multicast TV channel), > > how much trouble would a typical consumer router run into if we turn on > > multicast? > > >From my understanding most customer routers do not have the > multicast->wifi-unicast conversion mechanism, so it'd start to transmit > the multicast stream at low speeds without any retransmit mechanism, > giving that receiver lots of packet loss for this multicast plus chewing > up most of the airtime. > > So that's a bad idea... > > -- > Mikael Abrahamsson email: [email protected] _______________________________________________ MBONED mailing list [email protected] https://www.ietf.org/mailman/listinfo/mboned _______________________________________________ MBONED mailing list [email protected] https://www.ietf.org/mailman/listinfo/mboned