- GROW 75 - ARN francois - limited scope announcements draft-francois-limited-scope-specifics-00 See slides - show control/data plane inconsistency Permitted community tagging on routes from customer -> provider to limit path advertisements (Long set of demonstrations of communities being used to control route propogation) Start playing with the communities to shift traffic according to internal policies Data plane violations: Using some current methods to identify and alarm on this policy Look at netflow and alarm when traffic is seen where it ought not be Control plane violations: Seeing more specifics from peers they should not appear from (essentially use bgp collection to verify that the policy is being followed) Reaction methods for the violations: DAC - Delivery/Account/Charge Take evidence back to the violators, force charges/charge-backs This looks like violating each customer's policies DROP - acl traffic out, drop violations Are automated solutions to this proper? What other options exist? Comments please. jscudder - how much of a problem is this really? - there are simple configs, which impose policy, and which can be violated, operators need to take care for this Raszuk - Diverse Peering BGP draft-raszuk-diverse-bgp-path-dist-00.txt See slides - Essentially, send best-path on primary ibgp session then send 'other best path' on secondary ibgp session Polling GROW for acceptance/interest/reading Potentially need to identify the problem before this proposal Scudder - draft-ietf-grow-bmp-02 BMP update - Refresher of BMP's use-case 01 - last set of changes, fixed codepoints to match current deployment 02 - included draft comments from WG added messages for Peer-Up added l(ocal-rib) flag to header (not everyone can send adj-rib-in) version number editorial changes Feedback on 02 - vendor specific code-points - question to the audience? Is this something to do? NextSteps - Need more monitoring stations coded Implement changes from -02 to -03 'Are there other vendor implementations?' 'Nope, not known to the author' Lixia - Study on BGP duplicate messages See slides - this was covered at IEPG as well (same slide deck) Showing stats of particular AS's sending high numbers of duplicate updates Showing stats by breakdown of AS type/size/classification of duplicate sending Causes of route duplicity, RR's stripping attribures and reannouncing, supression of duplicate paths doesn't happen at the provider->customer edge Is this important? What could be done to find the caues(s)? Lixia - BGP Route collection session/collector stability slides - For longitudinal studies, there is no sense/collection of outages of sessions nor of collectors Software solutions are causing some of these outages? (quagga failures) Sales job for BGPMon tool lixia's students are developing. Paul Francis - Virtual Aggregation (see list of draft-ietf-grow-va-* drafts) slides - Next-Steps Autoconfig of/for VA MPLS scaling issues Writeup some deployment scenarios Overview of VA - Reduce FIB size without affecting RIB size claimed 10x reduction with limited effect of stretch VA requires as-wide implementation, to avoid loops limit FIB, push over tunnels to other devices which do have your known route Cover some sample deployments Outlined sample deployments...