[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

ISIS and OSPF together


Yes, if you are referring to RFC5838 like functionality in OSPFv3 (AF
support) that is correct.  I personally don't have experience with that mode
of operation (as the networks I had experience with went dual stack a while

I guess someone looking to dual stack now may want to consider that option.

I am personally biased towards IS-IS when looking to do both, but to each
their own.

To further my early points (not saying it's a good option, but adding some
context).  The rationale for keeping OSPFv2 was due to legacy tools and
operational procedures.  Adding a second IGP (years ago) for IPv6 was
considered (to some) a way of not specifically impacting the "bread and
butter" IPv4 service while turning up IPv6.

I guess all of that reasoning has likely changed for new IPv6 turn-ups as
there is much more operational experience with running multiple AFs now.

I should have highlighted the context before ? sorry.


Victor K

From:  Glen Kent <glen.kent at gmail.com>
Date:  Mon, 13 May 2013 00:13:38 +0530
To:  Victor Kuarsingh <victor at jvknet.com>
Cc:  "nanog at nanog.org" <nanog at nanog.org>
Subject:  Re: ISIS and OSPF together


Folks could, at least theoretically, use ISIS or OSPF multi instance/multi
topology extensions to support IPv4 and IPv6 topologies. This way they would
only need to run a single protocol and thereby requiring expertise in
handling only one protocol.

With whatever i remember, OSPFv3 can be used to support IPv4 as well - so
folks could also use OSPFv3 when they want to support both IPv4 and IPv6.


On Sun, May 12, 2013 at 6:17 PM, Victor Kuarsingh <victor at jvknet.com> wrote:
> Glen,
> One transition scenario you noted below is often a use case.  I have seen
> networks move from OSPF to IS-IS (more cases then the reverse).
> In those cases, the overlap period may not be very short (years vs.
> weeks/months).
> I have also seen some use one protocol (which I think was mentioned in
> another response) used for IPv4 and another used for IPv6.  The cases I am
> familiar, tended to be IPv6 with IS-IS and IPv4 with OSPFv2.
> I guess the reasoning here was that if you are running dual stack, with
> OSPF you will need to run two protocols anyway, so running OSPFv2(IPv3)
> and OSPFv3(IPv6) may not be that different then running OSPFv2(IPv4) with
> IS-IS(IPv6).  This dual stack option has run longer or is semi-permanent
> at times.
> A sub-case to the above may also be that one (operator) may want to
> leverage some of capabilities of IS-IS and may not be willing to get off
> OSPF for some reason.  The Multi-topology option in IS-IS may be quite
> useful if you have some functions which are non-congruent in your network
> and you want to maintain topology variations (multicast being one, or
> in-band management which I believe was alluded to in your OOB use case)
> Regards,
> Victor K
> On 2013-05-12 4:41 AM, "Glen Kent" <glen.kent at gmail.com> wrote:
>> >Hi,
>> >
>> >I would like to understand the scenarios wherein the service
>> >provider/network admin might run both ISIS and OSPF together inside their
>> >network. Is this something that really happens out there?
>> >
>> >One scenario that i can think of when somebody might run the 2 protocols
>> >ISIS and OSPF together for a brief period is when the admin is migrating
>> >from one IGP to the other. This, i understand never happens in steady
>> >state. The only time this can happen is if an AS gets merged into another
>> >AS (due to mergers and acquisitions) and the two ASes happen to run ISIS
>> >and OSPF respectively. In such instances, there is a brief period when two
>> >protocols might run together before one gets turned off and there is only
>> >one left.
>> >
>> >The other instance would be when say OSPF is used to manage the OOB
>> >network
>> >and the ISIS is used for network reachability.
>> >
>> >Is there any other scenario?
>> >
>> >Glen