>Chuck, I think this is a good question. I always looked for comparisons
>between IS-IS and OSPF and never really could find any good source (I mean,
>Doyle describe both protocols very well, but that's not what I'm looking
>for, I'm looking for large implementation descriptions, explaining
>problems/beneficts of using each one and experiences associated with it).
>
>UUNet for example uses IS-IS on their core while the rest (or the majority)
>of the ISPs use OSPF. I wonder why UUNet chosed for IS-IS instead of OSPF.
>Maybe somebody on the list has an answer?
I'm not sure if I'd say either protocol has a majority, depending on
how you define majority. There's probably a larger total number of
providers using OSPF as an IGP, but most (not all) of the largest
providers use ISIS.
Remember that providers do not use any IGP throughout their entire
network; the role of the IGP is more fast failure discovery within
subdomains, with the overall structure through carefully scaled iBGP.
Route reflectors are the most common iBGP scalability method, but
confederations do have their advocates (and applications).
Andrew Partan selected IS-IS for Alternet, the predecessor to UUnet.
At the time, there were two principal reasons:
--Cisco's IS-IS was more reliable than its OSPF code. OSPF had a major
rewrite and improvement in a late release of 9.1, but by then IS-IS
was operational.
--At the time of this decision (early 1990's), it was by no means certain
that the IP protocol suite would become dominant; there was still
significant interest in OSI CLNP.
>
>Today I would chose OSPF over IS-IS because:
>
>- much more vendors support OSPF compared to IS-IS
>- it's my perception that OSPF is the direction chosen by IETF, or at least
>much more work is being done on OSPF than IS-IS
>- OSPF is much more used than IS-IS, what makes easier to find people to
>implement/support it
:-) there's always the observation that Dave Katz wrote or modified
almost all the major research and vendor implementations of IS-IS, so
it's at least internally consistent.
>
>Any inputs on that?
>
>Guilherme
ISIS certainly is alive and well in many large providers, if for no
other reason that many routing architects for large providers started
their experience with Andrew at Alternet or Sean Doran at Sprint. As
they moved to other jobs, they stayed with what they knew worked.
At the same time, I am aware of major carriers that use OSPF quite
successfully. In most carriers, ISIS runs as a single area, even
with 1000 routers. OSPF has richer functions but be more complex.
There certainly is aggressive enhancement in the IETF for both OSPF
and ISIS. As mentioned in another posting, both are getting
facilities for traffic engineering.
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
