Hi Job, Thanks for the input regarding draft-keyur-idr-bgp-prefix-limit-orf-03.
That’s what I meant. Incoming peer prefix limit needs to be advertised by BGP. I think it’s a bad idea to configure statically an outbound peer maximum prefix limit. The only reason why I want that is to align with my peer. Ideally, these parameters should be automatically exchanged between peers. John and Robert point out that both of them are in competition which should take precedence. This points usually to a bad design. Let me go to the big picture and share my thoughts. The reason why we want inbound prefix limits is because we don't have endless resources in the BGP RIB's. Right? Thanks to https://tools.ietf.org/html/rfc7854#section-4.8 we are able to monitor their usage. Great. But we don't have any IETF YANG model available giving us the maximum possible route count on each BGP RIB. Why? Therefore we don't know at which threshold we should alarm. Nor we don't know at which static threshold we should limit the amount of prefixes we receive from which peer. Nor thus this mechanism account for the BGP local RIB maximum capable routes. If my BGP local RIB supports 1'000'000 routes and I have 10 peers with an average of 70'000 routes each. How high should I set the inbound prefix limit? My answer: +10% more prefixes in BGP local RIB than alert, +20% more in BGP local RIB than limit the peer which caused the issue and alert. Should be monitored with BMP stats reports (https://tools.ietf.org/html/rfc7854#section-4.8) and orchestrated automatically within BGP YANG model (container prefix-limit). Monitored and orchestrated in a closed loop operation. Do we want' that logic to be built in routers? Does one router has the big picture of the network and their peerings? No. Therefore it needs to be solved outside the BGP router. Kind regards Thomas Graf ____________________________________________________________________________ Network Engineer Datacenter Functions Telefon +41-58-223 84 01 Mobile +41-79-728 80 12 [email protected]<mailto:[email protected]> ____________________________________________________________________________ Swisscom (Schweiz) AG IT, Network & Infrastructure Datacenter Functions Binzring 17 8045 Zürich www.swisscom.com Postadresse: Binzring 17 8045 Zürich From: GROW <[email protected]> On Behalf Of Robert Raszuk Sent: Tuesday, March 26, 2019 5:02 PM To: John Scudder <[email protected]> Cc: [email protected] Subject: Re: [GROW] Prefix limit ORF And with this in mind the draft needs to define expected behaviour when locally configured outbound prefix limit is different from ORF pushed one (someone's inbound). Lowest wins, static wins, latest wins etc ... otherwise each implementation may choose differently ;) thx R. On Tue, Mar 26, 2019, 16:59 John Scudder <[email protected]<mailto:[email protected]>> wrote: Apropos Job’s talk just now — draft-keyur-idr-bgp-prefix-limit-orf-03 https://tools.ietf.org/html/draft-keyur-idr-bgp-prefix-limit-orf-03 —John _______________________________________________ GROW mailing list [email protected]<mailto:[email protected]> https://www.ietf.org/mailman/listinfo/grow
smime.p7s
Description: S/MIME Cryptographic Signature
_______________________________________________ GROW mailing list [email protected] https://www.ietf.org/mailman/listinfo/grow
