Hi Christopher, Where does it no longer make sense to deaggregate? Isn't that a bunch > related to what problem the initial announcement is trying to solve? >
My answer is - If routing path to destination by more specific would be identical to the path covered by less specific prefix. Based on some work on this in the past I do believe such "sweet spots" can be identified especially with average AS_PATH length being less then 4. Of course PA space is much easier then PI but both are possible to be aggregated. Would it cover all cases - perhaps no. Would it cover a lot of cases - likely yes. But before we spend more time on this let's answer one question which was asked before - do we have a problem ? Of course 1 mln routes is easier to be handled then 5 mln, but then you may stop upgrading hardware :) Not a colorful future for those who need to invest in their software ... Hi Job, > We need to document the cases where the business intentions and the reality of what happens in the internet routing system don’t match up. Fully agree and volunteer to help. Also in many cases we may indeed find that deaggregation is purely accidental ... however those who cause it just have no control to aggregate it or suppress announcements N AS hops away. So we could start picking low hanging fruits like Tony's draft-ietf-idr-as-pathlimit-03. Thx, R.
_______________________________________________ GROW mailing list GROW@ietf.org https://www.ietf.org/mailman/listinfo/grow