[83070] in North American Network Operators' Group
Re: Traffic to our customer's address(126.0.0.0/8) seems blocked by pa
daemon@ATHENA.MIT.EDU (Fergie (Paul Ferguson))
Wed Aug 3 23:50:31 2005
From: "Fergie (Paul Ferguson)" <fergdawg@netzero.net>
Date: Thu, 4 Aug 2005 03:48:56 GMT
To: pfs@cisco.com
Cc: nanog@merit.edu
Errors-To: owner-nanog@merit.edu
Mea culpa: I meant "a few /16's" as opposed to "2"...
No flames, it's too late...
- ferg
-- "Fergie (Paul Ferguson)" <fergdawg@netzero.net> wrote:
Philip,
This sounds very much like a bully -- 2 /16's are a major
problem, as opposed to a single /8?
Where is the major heartburn in this particlualr case?
I could understand if here were lots of farctured
annnounced space (granted: I haven't checked this yet),
but what's up with that?
- ferg
-- Philip Smith <pfs@cisco.com> wrote:
mkawano@bb.softbank.co.jp said the following on 4/8/05 12:03:
FWIW, if you don't announce your aggregate, do not be surprised if you
experience continued disconnectivity to many parts of the Internet. Some
SPs notice that SoftbankBB have received 126/8, so will likely filter as
such. Leaking sub-prefixes may be fine for traffic engineering, but this
generally only works best if you include a covering aggregate.
Try including your /8 announcement and see if this improves reachability
for you.
Out of curiosity, why pick on a /16 for traffic engineering? Most people
tend to analyse traffic flows and pick the appropriate address space
size as a subdivision. Or do you have 256 links to upstream ISPs and
need that level of fine-tuning?
best wishes,
philip
