Hi Shane,
Very belated (sorry!) answers to your questions in-line below.
On Nov 19, 2008, at 12:09 AM, Shane Amante wrote:
A few questions:
1) The draft says the following in Section 2:
---snip---
If the peer is a "Global Instance Peer", this field is zero filled.
If the peer is a "L3VPN Instance Peer", it is set to the route
distinguisher of the particular L3VPN instance that the peer belongs
to.
---snip---
I'm a little confused by the second sentence. Is that referring to
the case where a router is doing RFC 4364 "Option A" style eBGP with
a remote peer, or something else?
Basically, this means "anything that terminates on a VRF", which
includes "Option A".
2) In Section 3, the draft talks about sources of routing
information (Adj-RIB-In or Loc-RIB) sent in Route Monitoring
messages. Would it be useful to have a bit in the BMP "Peer Flags"
that indicates if the path being sent in a RM message was retrieved
pre- (Adj-RIB-In) or post- (Loc-RIB) policies, so the receiver knows
which he/she is looking at?
The "L" flag will be returning in the forthcoming revision.
3) How will BMP cope with BGP flap dampening being enabled on a BMP
source router? In other words, a router (configured as a BMP
source) receives a series of WITHDRAW's & UPDATE's that it is
configured to apply flap dampening on, suppressing re-advertisement
of these updates further into an AS. If the BMP source is
forwarding messages from Adj-RIB-In, would a BMP receiver see all
incoming WITHDRAW's & UPDATE's associated with a "flap" event, (even
though these updates would have been suppressed from further re-
advertisement after flap dampening is applied)? How would, or
should, a BMP receiver know that flap dampening is enabled and/or
would be applied?
Yes, all the flaps would be seen. (Modulo the possibility of state
compression eliding some messages, but the point is dampening wouldn't
be applied.)
In terms of the monitor knowing if dampening is enabled, conveying the
configuration of the router is well beyond the scope of BMP. I think
the answer is "look at the router config". In terms of the monitor
knowing if dampening was applied to a particular route, one can think
of heuristics to work this out by correlating BMP-learned data with
data from a regular BGP session.
The reason I ask is perhaps this is more a question of the
'authenticity' of "monitoring data" at a BMP receiver with respect
to formulating what did (or could) have happened with a series of
Updates being propagated further within an AS? Or, perhaps there's
another answer?
The conceptual model is to come as close as feasible to having the
monitor see what the router sees arriving from its peers. That's the
type of 'authenticity' we're going for.
HTH,
--John
_______________________________________________
GROW mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/grow
