Episode 19 – BGP: Traffic Engineering
UnderlayJanuary 10, 201849:4646.17 MB

Episode 19 – BGP: Traffic Engineering

In this Community Roundtable episode, returning guests Russ White and Nick Russo continue our three part deep dive into the Border Gateway Protocol, or BGP, with a look at the mechanisms within the protocol to perform traffic engineering.

Show Notes

Influence Ingress

  • Classic bestpath options to influence ingress
  • AS-path prepend outbound to influence inbound traffic
    • Why AS Path prepend doesn’t always work
      • In many areas, ISPs are in a full or almost full mesh and connected to common backbones making AS Path prepend largely irrelevant
      • Providers normally use their own local preference for outbound traffic back to a customer
    • MED
      • MED is a hint, it’s often stripped or ignored
      • MED only works if the AS Path is the same on all routes
      • MED is non-transitive and doesn’t mean anything beyond the next hop
      • Longest Match
        • Be careful about this, as it pollutes the DFZ
          • DFZ = default free zone
            • A router belongs to the DFZ if it doesn’t need a 0.0.0.0 route to reach everything on the internet
        • Tragedy of the commons here
          • An enterprise can force inbound traffic to be load-balanced better but it pushes the processing of that traffic engineering onto the internet
        • This is the “big hammer”
      • Using RFC 1998 communities for influence ingress traffic
        • This is a way to signal your provider to take some sort of BGP action
        • You need to find the specific communities used by each provider
        • Make certain the provider accepts communities on their eBGP edge

Influence egress

  • Local Pref
    • Overrides pretty much everything other than weight
    • Used to implement hot/cold potato routing
      • hot potato routing is when a provider chooses to get the traffic out of its network as quickly as possible at the closest egress point
      • cold potato routing is when a provider chooses to control some traffic as long as possible for some reason
  •  Weight
    • Local to a device
  • Other handy stuff:
    • Cost community: IGP and pre-bestpath POI
    • Accumulated IGP (AIGP)
    • iBGP tie breakers
  • Using RFC 1998 communities for influence ingress traffic
    • This is a way to signal your provider to take some sort of BGP action
    • You need to find the specific communities used by each provider
    • Make certain the provider accepts communities on their eBGP edge
  • BGP deterministic MED

 

Russ White Guest Nicholas Russo Guest Jordan Martin Co-Host Eyvonne Sharp Co-Host Phil Gervasi Co-Host

Outro Music: Danger Storm Kevin MacLeod (incompetech.com) Licensed under Creative Commons: By Attribution 3.0 License http://creativecommons.org/licenses/by/3.0/

The post Episode 19 – BGP: Traffic Engineering appeared first on Network Collective.