Hi folks,
Can anyone tell me about a situation in which a route which was not
valley free was not a result of a misconfiguration or a bad actor? For
those who don't recall the terminology, a network path is valley free
if it crosses exactly zero or one free peering links when traveling
between the two endpoints.
Thanks,
Bill Herrin
Assume 3 providers A B and C, where you have a single-homed customer on A and a
single-homed customer on C, and A and C don't peer. Traffic may end up going
thorugh an A-B peering and a B-C peering. And whether A-B and B-C are a free
peering or a paid transit is a business deal, outside the scope of BGP, unless
you want to abuse communities...
Are A and/or C "bad actors" for not peering? Jury is still out on that one.
Hi Valdis,
It's that business deal I want to hear about. When A-B and B-C are
free peering but the traffic goes A-B-C for some reason other than a
misconfiguration or deliberate abuse. On or off list, I'd like to know
about real-life use cases where folks do this on purpose.
Regards,
Bill Herrin
I can't think of any circumstances where the business "B" would be content transit traffic between A and C without some form of compensation. That compensation may not involve payment for bits, however. In theory, the compensation might be derived from something occurring at the application layer, but even in those cases that business relationship is probably not apparent from looking at prefix advertisements. Business B is probably using b2b user agents, gre encap or some other method that makes both legs look like independent IP flows to network A and B.
Interesting question, though.
Dave
The AS I worked at back in the day did to a degree for willing parties.
Mostly small ISPs who all knew each other. We had at the time 3 regional
hub locations with interlinks, and peered settlement free with 2 - 3 ASs in
1 of the locations, and 1-2 ASs each in the other 2 locations, all of which
could opt to allow their prefixes to be heard by the others via communities.
-Blake
If ASN B is a cooperative venture (such as a regional network) funded by A, C,
and several others for mutual gain, it's not at all out of the question.
I have worked for the middle network when I was responsible for a
government network - typically we were the middle network. Logic was it
was good for citizens for us to essentially act like a peering exchange for
certain types of entity (who also typically were government affiliated).
One I can think of was to allow a full mesh of video education between
various institutions - it was the right thing to do for all entities
involved and I facilitated it through the network I was affiliated with.
You might also think about the circumstance of two government
subcontractors working on a common project or interfacing with each other's
systems on behalf of a common customer. The middle network is paying each
end to connect to the middle but is providing reverse transit between them
(I.E. the end entities are paid to transit the middle!), although the
contracts aren't exactly phrased to say that! A lot of time, this may be
done with static routes, but it could easily be done with BGP and the end
effect is the same.
I have never heard the term valley free. Where does it come from?
It's that business deal I want to hear about. When A-B and B-C are
free peering but the traffic goes A-B-C for some reason other than a
misconfiguration or deliberate abuse. On or off list, I'd like to know
about real-life use cases where folks do this on purpose.
As far as I understand some NRENs do that in Europe. Check out AS1853
and AS-ACONETTOVIX in the RIPE whois. "A" networks are the peers a VIX,
"B" is ACONET, "C" networks are CESNET, SANET, and PIONIER.
DTAG's looking glass shows this path to SANET:
sh ip bgp regexp _2607_
BGP table version is 0, local router ID is 217.239.38.165
Status codes: s suppressed, d damped, h history, * valid, > best, i -
internal,
r RIB-failure, S Stale, R Removed
Origin codes: i - IGP, e - EGP, ? - incomplete
Network Next Hop Metric LocPrf Weight Path
*>i147.175.0.0 194.25.5.150 100 0 1853 2607 i
*>i147.213.0.0 194.25.5.150 100 0 1853 2607 i
*>i147.232.0.0 194.25.5.150 100 0 1853 2607 i
*>i158.193.0.0 194.25.5.150 100 0 1853 2607 i
*>i158.195.0.0 194.25.5.150 100 0 1853 2607 i
*>i158.197.0.0 194.25.5.150 100 0 1853 2607 i
*>i192.108.130.0 194.25.5.150 100 0 1853 2607 i
*>i192.108.131.0 194.25.5.150 100 0 1853 2607 i
*>i192.108.132.0/23 194.25.5.150 100 0 1853 2607 i
*>i192.108.138.0 194.25.5.150 100 0 1853 2607 i
*>i192.108.149.0 194.25.5.150 100 0 1853 2607 i
*>i193.87.0.0/16 194.25.5.150 100 0 1853 2607 i
*>i194.1.0.0/17 194.25.5.150 100 0 1853 2607 i
*>i194.160.0.0/16 194.25.5.150 100 0 1853 2607 i
Total number of prefixes 14
Regards,
András
This paper, which is a must-read for anyone interested in BGP:
Stable internet routing without global coordination
By Lixin Gao and Jennifer Rexford
http://dl.acm.org/citation.cfm?id=504612
once upon a time, provider A and provider P were having a peering war,
and provider V provided valley transit for P's prefixes to A. it was
not meant to be seen publicly, but the traceroutes were posted to nanog,
or maybe it was com-priv at the time.
this is far from the only time this has happened.
randy
Hi folks,
Can anyone tell me about a situation in which a route which was not
valley free was not a result of a misconfiguration or a bad actor? For
those who don't recall the terminology, a network path is valley free
if it crosses exactly zero or one free peering links when traveling
between the two endpoints.
Isn't that the way most of the IPv6 internet ran
for many years? ISP A -> 6939 <- ISP B,
settlement-free connections all around? It's
what established 6939 as the core of the
IPv6 internet.
Matt
Hi Matthew,
By peering I mean a link on which the two participants offer and
accept substantially fewer routes than "the rest of the Internet."
Usually only the routes for each participant's respective customers.
The clever folks at HE provided full IPv6 transit as a loss leader
which enhanced their market position (put them on the map quite
frankly). That's not a "valley" in this context.
I'm really intrigued by the multiple reports of RENs creating a sort
of shadow network where other RENs are permitted to cross their
internal backbone at no cost but not access their general Internet
transit. That does seem to be a valley. Is anybody outside the
Research and Education industry doing this sort of thing?
Regards,
Bill Herrin
Having been employed by a provider V in one such example of the below, I viewed it as a temporary, partial transit relationship. Does such a situation meet Bill's original definition?
Hi David,
I think you have the right of it. That the recipient elects only to
use the link for a limited set of destinations is an ordinary part of
transit service. In Randy's example, a peering link was converted to a
transit link on a short term basis.
Regards,
Bill Herrin
I think you have the right of it. That the recipient elects only to
use the link for a limited set of destinations is an ordinary part of
transit service. In Randy's example, a peering link was converted to a
transit link on a short term basis.
you know the term?
Hi Randy,
For my interests I don't care about the duration. Five minutes or five
years it's all the same to me. I care about the characteristics of the
relationship while it's ongoing.
Regards,
Bill Herrin