BGP for enterprise deployment

General
1

We’re building out two colos with a Balance 2500EC in each. Before I finalize the design I need to know what the 2500EC actually supports for BGP on 8.6.1. Mainly, I need to know if it can be our BGP edge to transit or if I have to put a router in front of it.

What I found on the forum is a few years old, so I’d rather get up-to-date info here. Yes/no is fine for most of these.

  1. allowas-in per neighbor, with a count. AWS originates our hub prefixes with the same ASN the colos will use, so without this the colos drop those routes.
  2. Setting communities on routes we announce, standard and large (RFC 8092). This 2020 thread made it sound like only received no-export is honored: BGP Community Support please
  3. Matching on received communities in an inbound policy.
  4. AS-path prepend per neighbor. In this thread, turning on prepend toward one neighbor stopped the route going to that neighbor at all: Multi eBGP peers Is that fixed?
  5. Per-neighbor inbound/outbound prefix lists with ge/le, AS-path filters, and setting local-pref and MED.
  6. RPKI origin validation (RTR to a validator). this isn’t a requirement but it’s how we’re doing it with cisco today.
  7. Max-prefix per neighbor.
  8. BFD on BGP sessions.
  9. Originating an aggregate anchored to a null/discard route.
  10. 4-byte ASNs.

Thanks in advance!

@James_Webster5240 tagging you because you’re the BGP king from what I hear :slight_smile:

2

Hi Chris

hope you are well, I have answered as far as I know but when advertising to the internet I have stuck to using another router for this and then just have the peplink device advertise the sites as required.
I’ve answered best I can in below:

  1. allowas-in per neighbor, with a count. AWS originates our hub prefixes with the same ASN the colos will use, so without this the colos drop those routes.
  • I don’t believe there is support for this.
  1. Setting communities on routes we announce, standard and large (RFC 8092). This 2020 thread made it sound like only received no-export is honored: BGP Community Support please
  • there is support for standard communities but not large.

  1. Matching on received communities in an inbound policy.
    -no only import prefix filters

  2. AS-path prepend per neighbor. In this thread, turning on prepend toward one neighbor stopped the route going to that neighbor at all: Multi eBGP peers Is that fixed?

  • I’ve done a couple of configs using AS-prepend from the peplink side and haven’t seen the issue mentions. Last one was on a pair of 2500’s 2 years ago and it worked fine.
  1. Per-neighbor inbound/outbound prefix lists with ge/le, AS-path filters, and setting local-pref and MED.
  • you have import list with an exact option, so you can either allow/disallow a prefix exact or LE the prefix
    there is no GE/LE, no AS filters or MED.

  1. RPKI origin validation (RTR to a validator). this isn’t a requirement but it’s how we’re doing it with cisco today.
    -No

  2. Max-prefix per neighbor.
    -No

  3. BFD on BGP sessions.

  • yes and it has worked fine in the instances where I have used it.

  1. Originating an aggregate anchored to a null/discard route.
    -there is no direct blackhole or null routes, I guess you could add a large prefix on the lan as an additional subnet and then let the more specific routes get priority but it would be a bit messy. OSPF has the custom route advertising and that might work for BGP as well but I haven’t tested it.

  2. 4-byte ASNs.
    -not sure as I haven’t tried it but no lage community support means no 4 byte ASN in the community tags

thanks
James