[rbridge] [Isis-wg] Some comments about TRILL ESADI processing of CSNPs

[Radia Perlman]

The possibilities are:

a) don't worry about it -- it doesn't seem like a major problem if
occasionally multiple RBridges send CSNPs. Maybe add explicit
instructions for jitter so all RBridges don't time out at the same
time and send CSNPs.

b) have an RBridge receiving CSNPs defer to a higher priority RBridge
sending CSNPs, in the sense of ceasing to send CSNPs if it gets even a
single CSNP fragment from a higher priority RB.

I think there will be another document describing the details of
ESADI, in which case we can make those clarifications there.

Radia



On Tue, May 11, 2010 at 4:02 AM, mike shand <mshand at cisco.com> wrote:
> I'm sorry, I have only just noticed the following
>
>
> In section 4.2.5.1 of the TRILL spec it says
>
> "  The ESADI DRB sends TRILL-ESADI-CSNP frames on the ESADI virtual
>   link. *For robustness, a participating RBridge that determines that
>   some other RBridge should be ESADI DRB on such a virtual link but has
>   not received or sent a TRILL-ESADI-CSNP in at least the ESADI DRB
>   holding time MAY also send a TRILL-ESADI-CSNP on the virtual link.* A
>   participating RBridge that determines that no other RBridges are
>   participating in the ESADI protocol for a particular VLAN SHOULD NOT
>   send ESADI information or TRILL-ESADI-CSNPs on the virtual link for
>   that VLAN.
> "
>
> However, ISO/IEC 10589 second edition in clause 7.3.17 says
>
> If an Intermediate system detects that the transmitter had more up to date
> information, the receiving Intermediate system shall multicast a Partial
> Sequence Numbers PDU (PSNP), containing information about LSPs for which it
> has older information. This serves as an implicit request for the missing
> information. Although the PSNP is multicast, *only the Designated
> Intermediate System of the appropriate level shall respond to the PSNP.*
> NOTE 38 This is equivalent to the PSNP being transmitted directly to the
> Designated Intermediate System, in that it avoids each Intermediate System
> unnecessarily sending the same LSP(s) in response. However, it has the
> advantage of preserving the property that all routeing messages can be
> received on the multi-destination addresses, and hence by a LAN adapter
> dedicated to the multi-destination address.
>
>
> Presumably in the ESADI case, the intention is that the non-DRB sending the
> CSNP should ALSO respond to the PSNP.
>
> It probably therefore needs some words to clarify this.
>
> However, I am more concerned by the following.
>
> Presumably the intention is that the first non-DRB to notice the lack of
> CSNPs and send one itself will inhibit other RBridges from also sending
> CSNPs, hence avoiding a storm of CSNPs.  However it is possible that some
> number of other RBridges will start to transmit CSNPs before they receive
> the CSNP sent by the first RBridge. Since the CSNP will in most cases be a
> set of CSNPs, is the intention that reception of a CSNP should inhibit the
> sending of further members of the set, or should the complete set be
> transmitted once it has been started, even in the presence of CSNPs from
> another RBridge?
>
> I think it has to be the latter, since two RBridges deciding to send a CSNP
> set at about the same time could mutually inhibit each other, resulting in
> only a partial CSNP set being transmitted by either of the RBridges, and
> thus negating the effect of the "backup" CSNP.
>
> There would appear to be some need of clarification here.
>
>
>
>    Mike
>
>
>
> --
> For corporate legal information go to:
> www.cisco.com/web/about/doing_business/legal/cri
>
> _______________________________________________
> Isis-wg mailing list
> Isis-wg at ietf.org
> https://www.ietf.org/mailman/listinfo/isis-wg
>