Re: RE: AD request / L2 Triggers Chapter Statement
Joe Touch <[email protected]>
| Newsgroups | gmane.ietf.pilc |
|---|---|
| Message-ID | <[email protected]> |
Phil Neumiller wrote: >>The whole of L2 Triggers assumes that: >> >>1) events occur at the link on the timescale of milliseconds >>2) these events must be communicated to L3 >>3) L3 can do something useful with these triggers >> >>#1 is granteed, but #2 assumes that #3 ever occurs. There is little in >>IP as a whole that supports such "realtime" requirements of millisecond >>responses. In fact, that's in opposition to most of the rest of the >>Internet architecture. > > > Yes. That IS the point. There needs to be. But there isn't; IP currently won't do anything interesting with info on that timescale. It's like calling up someone who won't answer. > If IP were to create a nice > home for L2s, life would be much simpler for wireless system providers. > Wireless links stink. They come and they go. IP was build for 10E0-06 > or better BER links. This needs to change. Yes - but, as I suggested later on, it can change by creating a wireless virtual link that goes over different wireless physical links, ala ATM. It need not involve IP. > It needs to change in IP not > in the L2s. The L2s do what they can. I would beg to differ that this > is NOT realtime. Many routers handle millions of packets per second > over optical links. THIS IS VERY REALTIME. Nothing about IP requires that packets be routed in microseconds. The RATE of routing is fast, but the latency need not be so low. Further, we're talking about updating routing, which is much slower in general. > Applications include but are not limited to: > > o Fault tolerant failover of links Usually within a single link layer type - i.e., from one 300Mhz wireless link to another 300Mhz wireless link. That's what I mean as "intra link" - the link layer could have handled that. > o L2 signal quality indications To what end? What would IP do differently? > o L2 congestion control Again, to what end? IP doesn't have congestion control; TCP doesn't react on milliseconds. > o L2 fragmentation control There's an existing mechanism for that - MTU. Does it change on the link on the fly? Joe _______________________________________________ pilc mailing list [email protected] https://www1.ietf.org/mailman/listinfo/pilc http://www.ietf.org/html.charters/pilc-charter.html http://pilc.grc.nasa.gov/