At this point of discussion, I think that Yakov's
"the architecture should *not* be prohibited from using
... services from the layer below" is the safest
formulation.
-- dima.> -----Original Message----- > From: Sean Doran [mailto:smd@ab.use.net] > Sent: Tuesday, March 26, 2002 3:15 PM > To: FKastenholz@unispherenetworks.com; yakov@juniper.net > Cc: dima@krioukov.net; irtf-rr@nether.net > Subject: Re: Additional comments on draft-rrg-kastenholz-req-04.txt > > > > | I certainly agree with you that the architecture must allow for > | "as wide a variety of media as possible". However, I also think > | that the architecture should *not* be prohibited from using > | routing/topology/policy services from the layer below. > > There is an obvious tension between using extra information that is > already provided by the layer(s) below, whether those are HDLC keepalives, > SDH/SONET B1/K1/K2, circuit AIS, ES:IS, ARP, or whatnot, and > *requiring* that the layer(s) below provide certain intelligence > information to the routing system. > > That said, is there *some* information that SHOULD/MUST be provided > by the layer(s) below to the routing system, or do we assume the > presence of completely unintelligent lower-layers, along the lines > of the very first SLIP connections using Rick Adams's drivers? > > Sean. > > PS - there is also some tension (even today) between making use of > a lower layer that has its own routing system, and migrating > control of those lower layers to "our" routing architecture/system.
This archive was generated by hypermail 2b29 : Mon Aug 04 2003 - 04:10:04 EDT