Last Modified: 2004-09-28
|Done||Hold the first Working Group meeting and discuss the charter and the state of progress on the chartered items.|
|Done||Update the PIM-DM version 2 Internet-draft. Go to WG last call. Submision to IESG for considerations as proposed standard.|
|Apr 04||Resubmit the latest PIM-SM version 2 specification to IESG for consideration as a proposed standard RFC|
|Aug 04||Submit PIM-SM Applicability Statements|
|Aug 04||Submit PIMv2 MIB to IESG for consideration as proposed standard.|
|Sep 04||Submit updated PIM-SM and PIM-DM internet-drafts, clarifying any inconsistencies or ambiguities in the previous drafts.|
|Nov 04||Submit PIM-SM version 2 and PIM-DM version 2 specifications to IESG for consideration as Draft Standards.|
|Mar 05||Submit PIMv2 MIB to IESG for consideration as draft standard.|
Thanks to Marshall for the minutes:
IETF 61 : PIM WG
Tom Pusateri and Mike McBride presiding
Status of the new standard of PIM-SM -v2
For a new standard need
1. A well written spec - absolutely
2. Must have an Internet draft of the MIB - exists, plus a new draft in the works
3. Security architecture in the spec (they exist)
4. At least one implementation : We have at least Infosys, IOX, Xorp.
Isidor Kouvelas : IOS is using it for IPv6
Mark Handley : We tried to make the spec backwards compatible as well, so
old implementations in a sense count too.
Tom Pusateri : But they were not written using the new spec, so they don't.
5. Major features tested. This is done using the deployed code
Mark Handley : for Xorp we have a pretty complete internal test suite.
6.) Write up and submission to the IESG.
This is in the works.
At that point, the document will be out of the WG.
Stig Venaas - email@example.com
BSR spec. Graphs are at
How to withdraw RPs from a group range
IWNH - Implicit withdrawal of the RP, by not including it on the list
EW Expiry - Explicit : use a hold time of zero to withdraw an RP
IW - combination of the above
Previous specs seemed to be IW or IWNH but not clear from the drafts.
Some backwards compatibility issues.
Which approach should the group back ? Each approach takes some real work, so we don't want to do ones that are not needed. We need feedback as to what implementations exist.
Mark Handley. The spec is badly done, so the intent does not matter. These seem to be equivalent in practice. What we need to know is what people are doing
Bharat from Infosystems : We use a hold time (EW) approach.
Paul Kepler of BG systems : We do implicit withdrawal, because that is how we read the spec.
Stig - We need to poll the group to see what the majority of people actually do.
Stig - How to cope with IPv6 and admin scopes
Scopes are numerically 1,2,5,8,E, etc.
Scope 5 gives FFc5::/16
Each scope has 16 possible ranges.
FF05::/16 FF15::/16 FF25::/16 etc.
Don't want too have to specify 16 different ranges.
Is a simple 4 bit scope Identifier sufficient ?
How should we encode it ?
There may be issues with scopes with overlapping ranges and unions of ranges.
Brian Habermann. I had a conversation with Steve Deering. The intent was that you should only look at the 4 bits and the flags are irrelevant.
I would argue that you would want to keep it simple. Just use the 4 bits.
Toerless - What keeps one scope from being ASM and another being Bi-Dir ? Is that allowed ? I would like to be able to do this.
Mike McBride : Some notes from the co-author : We want to keep the v4 and v6 specs the same. If we only use 4 bits then v4 and v6 will diverge.
Bharat Joshi, Infosys India
How many people want this done ?
(About a 40% raise of hands.)
John Meylor - Don't correlate the amount of feedback with the importance of the draft.
Original Pim MIB authorized by J. Nicholas - expired May 2004 Defined Sparse and Dense Mode tables
IPv6 needs to be done - there is a document in review. The tables are pretty similar with IPv4.
Bill Fenner - There was a proposed v6 MIB proposed, but so long ago that it is forgotten
Bharat - I looked for this but I couldn't find it. - can you send this to the list ?
Tom Pusateri - SHould it be referenced assuming it can be found ?
Fenner - I think that this should be independent of the IPMroute MIB. A loose association between the two tables is OK - the current tight association is not good.
Bharat : Actions :
Ran "smilint" on the MIB and fixed any problems made changes to make the order proper added a new object for interface BSR borders
Mark Handley : What is a BSR border ?
Bharat : It is a BSR interface at a scope border
Mark - But that is not a PIM thing. It maybe should be somewhere else.
Bharat - There may be changes because of changes in PIM-SM and BSR drafts.
Road Map :
Need more reviewers of the draft
Need an extensive review of he IPv6 MIB
The result will be PIM MIB v 2
There are separate IPv4 and v6 tables; these need to be combined.
PIM MIB mailing list is at
Bill Fenner - Can we combine this work back into the main PIM WG mailing list ?
Mike McBride - the following drafts are not yet part of the PIM WG. At MBONED we had discussion about expanding the PIM work to include MPLS LSP and related topics.
Who would be open to expanding the PIM charter ? (Maybe 20 hands or 30%)
Who would not like to do this (no hands at all).
Arjen Boers, Cisco Systems
IJsbrand Wijnands, Cisco Systems
How do you do PIM in networks with no routes towards the RP ?
Solution : Use a proxy. There needs to be more information to do routes back to the RP. Do this by adding TLV to the PIM join
Proxy types :
Isidor Kouvelas : Why are you calling this a proxy ?
Arjen : Because you can put a lot of things in the proxy.
2.) MDT-SAFI - for the interdomain or VPN case case
Vector plus RD is added to the proxy.
Can build an explicit path in the network
Isidor Kouvelas : Shouldn't the TLV's be defined in the draft ? Also, this means of carrying additional information is not the same as a proxy.
Yakov Richter - You want to explicitly route PIM join messages. What is the problem you are trying to solve ? The MDT-SAFI version is based on the Rosen draft, which was never approved by the L3VPN group.
How many providers would like to have a BGP free core ? That is what this is really aiming towards.
John Meylor : This is never adequately addressed in the working group. The providers have said that that they need it.
Mark Handley. At times there are circular dependencies, and at some times you just have to break them by moving on.
Mike McBride - the chairs will be meeting tomorrow to discuss these matters.
Arjen - We are not claiming to meet the requirements documents,
Yakov - we need the requirements before we consider solutions to those requirements.
Tom Pusateri - It might be useful to split this into 3 documents.
Isidor Kouvelas - I can see other applications here. This is a very basic building block.
John Meylor - There have been service providers that have stood up and expressed interest in the vector proxy draft.
Tom Pusateri - L3 and AT&T expressed interest on the list.
John Meylor - It is near miraculous to get 2 ISPs to discuss such interest on a mailing list.
Bill Fenner - I have heard that AT&T wants to remove BGP from the core to guard against possible denial of service attacks on the core.
Mike McBride - L3 said on the mailing list that not having this is a competitive disadvantage.
Toerless - this is a modular component which can be used or not. What sort of solutions requirement was requirement before IP source routing ? This is a pure control plain change.
Yakov - If you want to introduce this into PIM, found it on requirements from the requirements document.
John Meylor - One thing that this IETF has done is to expose the absolute confusion between some of the IETF working groups.
Mark Handley - What are the deployment concerns - do you have to upgrade all of your PIM routers in the domain ?
Arjen - Correct.
Two new drafts
This introduces accounting information into the PIM join packets.
Not really a completely new protocol - but should it be in PIM ?
James Leelow from Data Connection - We need a generic TLV mechanism to add information to the PIM Join. This should have a common format if it is going to be done at all.
Toerless - Which group is going to write a requirements document for this
Tom Pusateri - I do not think that this is productive. No one on the list has been arguing for this (the POP-Count draft). As for the proxy draft, Mike and I will be meeting with the other working group chairs
Toerless - I think that there is an abuse of process. Why should feedback from other WG always be required to pass judgment on a draft before it goes forward ? This can always be used to shoot down a draft.
John Meylor - Multipoint LSP is a tough nut to crack, so working with other working groups makes sense.
This draft makes it so that the anti-replay mechanism is not disabled, as is the case for the PIM-SM draft.
One comment on the list, from Pekka, said that link local threats to PIM should be added to MAGMA draft dealing with link-local drafts for IGMP and MLD.
Brian Habermann - Greg's draft for MAGMA is just a individual submission. This probably should not be added to that - this would be applicable to other routing protocols than PIM.
Tom Pusateri - At the L3VPN WG meeting, it was suggested to add state refresh reduction to PIM. This is an RFC for this for RSVP, so that might be used as a template.
Toerless - What else should be added to PIM once the proposed standard is published ?
Pusateri - When you start getting large amounts of (S,G) it starts to be an issue to refresh it all every 60 seconds.