Tony – My understanding is that algorithm definitions are topology independent. This is consistent with the two algorithms defined today (standard SPF and strict-spf). Using an algorithm in a particular topology would require topology/algorithm specific SIDs and the advertisement of topology specific link attributes (i.e., in an MT specific IS Neighbor).
Les From: Isis-wg [mailto:[email protected]] On Behalf Of Tony Przygienda Sent: Monday, November 20, 2017 1:11 PM To: Acee Lindem (acee) <[email protected]> Cc: [email protected]; [email protected] Subject: Re: [Isis-wg] ISIS SR Flexible Algorithm (Resending with alias correction) So, it seems that there will be a new draft with 242 covering all algorithms (i.e. no MT specific algo advertisement anymore). Then I thought each MT advertises which Flex it supports. Is the assumption that you can run multiple algorithms per MT? How would you otherwise have a two-algorithms-to-same-prefix problem? Is there some kind of conceptual model of FlexAlgo, i.e. how many of what associated with how many of the other (MT to algo, algo to protocol instance etc) ... -- tony On Mon, Nov 20, 2017 at 10:19 AM, Acee Lindem (acee) <[email protected]<mailto:[email protected]>> wrote: Hi Shraddha, Peter, et al, The comment on the draft I had was that the conflict case where two ISIS routers advertise the same multi-homed prefix with a different algorithm needs to be covered. I wouldn’t try and optimize for this and would just do whatever is simplest but avoids loops (e.g., log the situation and prefer the path computed with the lowest numbered algorithm). Thanks, Acee _______________________________________________ Isis-wg mailing list [email protected]<mailto:[email protected]> https://www.ietf.org/mailman/listinfo/isis-wg
_______________________________________________ Isis-wg mailing list [email protected] https://www.ietf.org/mailman/listinfo/isis-wg
