ietf
[Top] [All Lists]

RE: Simple question

2002-02-13 08:10:02
From: Michel Gilbert <m(_dot_)gilbert(_at_)hill(_dot_)com>

There will always be a brief period of time immediately following the
failure when R1 will blithely continue to forward packets oblivious to the
loss of the R4. How long that period of time will be depends on which
routing protocol the routers are implementing and how the protocols are
configured. "Modern" routing protocols (e.g., EIGRP, OSPF) can detect such a
failure in a matter of seconds. Older routing protocols (e.g., RIP) could
take a couple to a few minutes.

That's mistaken.  In the cases where EIGRP, OSPF, and other protocols
will detect and announce a dead link within a few seconds, RIP will also
do a flash-update within a few seconds.

Note that I intentionally wrote "RIP" instead of "RIPv2" and instead
of "RIP as implemented by very large router vendors" because RIP as
defined and implemented in the de facto standard and reference open
source implementation has had flash update and other features since
the 1980's and long before RIPv2.

Note that this entire thread does not belong in this the IETF's main
mailing list, except perhaps as an object lesson in the unavoidable
and major limitations of conformance testing or conformance certifying
and as a counter example to the previous thread.  In practice, no
matter or what who says, most users will assume that their chosen
vendors do it right.  When their chosen vendor is very large in the
market, their assumptions will approach religious certainty.


Vernon Schryver    vjs(_at_)rhyolite(_dot_)com



<Prev in Thread] Current Thread [Next in Thread>