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

Re: OSPFv3 Extension Future Direction (this ought to get your attention)

Pyda,
See inline.

Pyda Srisuresh wrote:

--- Acee Lindem <acee at CISCO.COM> wrote:


Pyda Srisuresh wrote:



--- Acee Lindem <acee at CISCO.COM> wrote:




Pyda Srisuresh wrote:





Acee,

I had done some work previously on using new LSAs for advertising topology




and




prefix info into TE networks (draft-srisuresh-ospf-te-07.txt). This draft




can




be a good starting point to use for turning into future MT extensions to
OSPFv3.






Hi Pyda,
I don't see a requirement to mix TE data and the unicast/multicast topology information. This has traditionally
been separate and processed by different functional components. The new LSAs in our draft correspond
directly to the existing base LSA types (only they are in extendible TLV format and support multiple
topologies).





[suresh] Acee - I am referign to usign the srisuresh-ospf-te draft as a


basis


to make MT extensions. Specifically, the suggestion is to add new LSA types
with MT protocol extensions. And, the new LSA types could have topology
specific TLVs. The LSAs may be restricted to being advertised within the
topology boundary.




Pyda,

I took another look at your draft and don't see a good reason to use it as a starting point
for OSPFv3 extension including MTR and AFs:

1. It is targeted at TE extensions
2. It is based on OSPFv2
3. It doesn't specify the changes needed to base OSPFv3 needed to support it.
4. To the best of my knowledge, there is not a large base of deployments or even
implementations that would benefit from using it as a base.




Acee,

Items 1, 2 and 3 are true. However, that is not to say that the concepts in the
draft cannot be used as the base to add new LSA types with MT protocol
extensions, restrict scope of LSA advertisements etc. The draft will need
rework and will look much different from the current draft when done. 

I suspect it would look much like draft-mirtorabi-mt-ospfv3-01.txt.

As for Item 4, you are postulating that if we did have a new draft using the
concepts in srisuresh-ospf-te, you dont think it woudl benefit the existing
deployments. That seems a bit presumptious. Sound like, you are declaring that
adding new LSAs will not benefit existing deployments and hence not beneficial
to pursue. If so, it seems you have already made up your mind to go with the
TLV based approach. Is that the case?


That is not the case at all. #4 was in direct response to the presumption that there was tremendous
advantage to be gained in starting with srisuresh-ospf-te as a base since it already exists.

As for the TLV approach, I do think it is the right way to go but it is the OSPF WG and not any
single individual that must decide. The OSPF WG is always open to other proposals.

Thanks,
Acee


cheers,
suresh



Thanks,
Acee



cheers,
suresh





Thanks,
Acee





I have not been active on this work recently due to my work priorities.
However, if there are folks on the list interested in pursuing, I would

be


happy to work with you on the effort and discussion. 

Thanks.

cheers,
suresh 

--- Acee Lindem <acee at CISCO.COM> wrote:






Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

At the Minneapolis IETF, I presented a comparision of  the
various options we have for extending OSPFv3. The reaction to the
presentation was very positive. However, it was last on the agenda and
the OSPF WG coincided with other WGs (most notably L2VPN).

At this point, I would like to initiate more discussion. The basic
question is whether we go with a TLV based approach or continue
to add new LSAs which need to be synchronized with the LSAs
advertising topology and prefix information. A TLV based approach
is described in:

http://www.ietf.org/internet-drafts/draft-mirtorabi-mt-ospfv3-01.txt

The crux of my presentation would be that this is gives us the powerful
extension capability while still allowing us to advertise information


that


does not need to be synchronized in separate LSAs.

Thanks,
Acee