The MPLS WG Archive[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index] Optical link bundling. Was Re: Draft Minutes From Pittsburgh
Yangguang, Wednesday, October 18, 2000, 11:05 AM, Yangguang Xu <xuyg@lucent.com> wrote: > Alex, > When using independent application to distribute optical topology, a TCP session > is enough. Got it. It was a more generic comment. > If we want to apply OSPF directly, we can set up a FA-LSP between two NEs that > are adjacent in their transport(data) plane but not adjacent in their control > plane. The purpose of FAs is exactly the opposite---announce an LSP as a topological entity without establishing a control adjacency over it. I believe you mean just an LSP. Thanks, Alex. > Either way, you don't need a dedicated physical control channel. > Thanks, > Yangguang > Alex Zinin wrote: >> >> Yangguang, >> >> > Alex, >> > I wasn't simplifying it. Indeed, I make it more complicated and generic. IGP and >> > signaling shouldn't make invalid assumptions either. Circuit switched transport >> > network is different from IP network. >> >> Do you mind explaining how you expect the two components to work over >> a TCP connection btw non-adjacent routers? >> >> Thanks, >> >> Alex. >> >> > Yangguang >> >> > Alex Zinin wrote: >> >> >> >> Yangguang, >> >> >> >> I wouldn't oversimplify this. >> >> Don't forget about IGPs and signaling. >> >> >> >> -- >> >> Alex Zinin >> >> >> >> Tuesday, October 17, 2000, 3:16 PM, Yangguang Xu <xuyg@lucent.com> wrote: >> >> >> >> > Hi, >> >> >> >> > This general notion is correct and it is against the very basic assumption of IP >> >> > network where control traffic is mixed together with data traffic. >> >> >> >> > In circuit switched network, there should be no assumption about the relation of >> >> > data plane topology and control plane topology at all. For two NEs to talk, you >> >> > don't need a dedicated control link. You only need a TCP session for a NE pair, >> >> > not only a whatever bundle. >> >> >> >> > Cheers, >> >> >> >> > Yangguang >> >> >> >> > Vishal.Sharma@tellabs.com wrote: >> >> >> >> >> >> Keereti, >> >> >> >> >> >> I believe what Frank is talking about is not merely the separation >> >> >> of the control channel(s) from the data channels, but rather a more >> >> >> general notion of non-associated out-of-band signaling. >> >> >> >> >> >> -Vishal >> >> >> >> >> >> > -----Original Message----- >> >> >> > From: kireeti@juniper.net [mailto:kireeti@juniper.net] >> >> >> > Sent: Tuesday, October 17, 2000 5:30 PM >> >> >> > To: fhujber@hotmail.com; mpls@UU.NET >> >> >> > Subject: Re: Draft Minutes From Pittsburgh >> >> >> > >> >> >> > >> >> >> > Hi Frank, >> >> >> > >> >> >> > > My perspective is naive, but as I understand it, the >> >> >> > current proposals >> >> >> > > require a control link to "overlay" an optical link >> >> >> > exactly, and for the >> >> >> > > control path(s) to be associated with groups of optical >> >> >> > links according to >> >> >> > > their characteristics (i.e. bandwidth - OC-48 vs OC-192). >> >> >> > >> >> >> > > Some of us in the optical world, growing from the telephony >> >> >> > (SONET/SDH) >> >> >> > > world do NOT assume that this overlay occurs. We prefer to offer the >> >> >> > > provider (in my case, my customer) to have management >> >> >> > networks that are not >> >> >> > > the same as the transport networks. This is because in-band >> >> >> > signaling is not >> >> >> > > yet available (though OIF is working on it) and because >> >> >> > providers may not >> >> >> > > want to dedicate a wavelength to relatively low-speed >> >> >> > management traffic. >> >> >> > >> >> >> > We have the notion of separate control channels and data paths in many >> >> >> > drafts. See draft-ietf-mpls-lmp-00.txt for a brief description. >> >> >> > >> >> >> > Kireeti. >> >> >> >
|
|