headers
David Ward | 2 May 2008 18:04
Picon
Favicon

WG LC on MPLS doc

All -
Many apologies for getting this draft to you late for review but, it is in IETF LC:



They authors, MPLS WG chairs and Routing ADs would like you to give comments in  a 2 week WG LC. Please send all comments on this draft by May 16, 2008.



In particular I want to call to your attention on LSP origination trigger mechanisms (which are stated outside the scope of the draft). It appears to be unclear what happens during a link flap with an interesting number of tunnels traversing that link. Will it trigger LSP flooding to advert/update the TE-LSP-count subtlv? Will this scenario cause a storm? RFC 3784 states with respect to sub-TLV 11:

" For stability reasons, rapid changes in the values in this sub-TLV
   SHOULD NOT cause rapid generation of LSPs."

How should this be addressed? What should the ISIS WG state as a general policy for other WGs to put in their docs as we have run into this issue several times in the last few years?

Many thanks.

-DWard, Chopps


_______________________________________________
Isis-wg mailing list
Isis-wg <at> ietf.org
https://www.ietf.org/mailman/listinfo/isis-wg
Permalink | Reply | 0 comments |
headers
Ilya Varlashkin | 2 May 2008 18:40

Is ATT bit mandatory?

Hi,

If L1/L2 router is configured to leak all necessary L2 routes into L1,
does it still have to set ATT bit? Assuming an implementation that
conforms to all relevant IS-IS standards/specifications, will something
fundamentally break if ATT bit is not set but there's clear evidence
that router must be L1/L2 (e.g. by presence of inter-area routes in its
LSP)?

I've made lab tests, and it appears I can turn ATT bit off on L1/L2
router and still have necessary connectivity (don't need default route
in L1), but I wonder if there are some hidden pitfals in implementing
this on life network (besides bugs).

Kind regards,
iLya
Permalink | Reply | 0 comments |
headers
David Ward | 9 May 2008 15:15
Picon
Favicon

Re: WG LC on MPLS doc

Hannes -

I agree. In the past there have been statements in drafts about the  
need to pace origination and it's overall impact. How often, what  
pacing algorithm are often not in scope of the spec in hand.

-DWard

On May 9, 2008, at 2:09 AM, Hannes Gredler wrote:

> hi david, et al,
>
> i do think that LSP origination triggers, and actual LSP origination
> are two seperate things. Any sensible link-state protocol  
> implementation
> should make sure to not issue more than one self-originated
> LSPs in N (e.g. 10) seconds.
>
> This is very similar to SPF triggering and actual SPF calculation.
> Not every single trigger causes an immediate calculation.
>
> So as far as standardization is concerned, i do not think that we  
> should
> worry as long as the spec-authors keep in mind that self-originated
> LSPs will be paced at a sensible rate.
>
> /hannes
>
> On Fri, May 02, 2008 at 11:04:11AM -0500, David Ward wrote:
> |    All -
> |
> |    Many apologies for getting this draft to you late for review  
> but, it
> |    is in IETF LC:
> |
> |    [1]http://tools.ietf.org/html/draft-ietf-mpls-number-0-bw-te- 
> lsps-09
> |
> |    They authors, MPLS WG chairs and Routing ADs would like you to  
> give
> |    comments in  a 2 week WG LC. Please send all comments on this  
> draft by
> |    May 16, 2008.
> |
> |    In particular I want to call to your attention on LSP origination
> |    trigger mechanisms (which are stated outside the scope of the  
> draft).
> |    It appears to be unclear what happens during a link flap with an
> |    interesting number of tunnels traversing that link. Will it  
> trigger
> |    LSP flooding to advert/update the TE-LSP-count subtlv? Will this
> |    scenario cause a storm? RFC 3784 states with respect to sub- 
> TLV 11:
> |
> |    " For stability reasons, rapid changes in the values in this  
> sub-TLV
> |
> |       SHOULD NOT cause rapid generation of LSPs."
> |
> |    How should this be addressed? What should the ISIS WG state as a
> |    general policy for other WGs to put in their docs as we have  
> run into
> |    this issue several times in the last few years?
> |
> |    Many thanks.
> |
> |    -DWard, Chopps
> |
> | References
> |
> |    1. http://tools.ietf.org/html/draft-ietf-mpls-number-0-bw-te- 
> lsps-09
>
> | _______________________________________________
> | Isis-wg mailing list
> | Isis-wg <at> ietf.org
> | https://www.ietf.org/mailman/listinfo/isis-wg
>
Permalink | Reply | 1 comment |

Gmane