a fellow researcher wants
> to make the case that in some scenarios it is very important for a
> network operator to be able to specify that traffic should *not*
> traverse a certain switch/link/group of switches/group of links
> (that's true right?). Could you give some examples? Perhaps point
> me to relevant references?
if so, why? security? congestion? other? but is it common? and, if
so, how do you do it?
randy
Hi Randy,
Depends on the context of the question. There's a simple concept a
surprising number of routing researchers don't fully grasp: we like to
be paid.
Scenario: a free peer and a paying customer can swap packets via my
links but two free peers may not. A free peer definitely should not
have access to the upstream transit links I have to buy. If nobody is
paying me for that packet, I'd like it to take the long way around.
Any way but through my network.
And yes, as you know it is very common for ISPs to strenuously
disapprove of unpaid transit. And we mainly do it by limiting the
propagation of free peer routes we received via BGP.
Seems like this should be so obvious as to need no mention. It's not.
Regards,
Bill Herrin
I don't think it is common, but I have a microwave network made up of a combination of license-free links and amateur radio band links (where no commercial traffic is permitted). For now the ham-band links are stubs, so that's easy. But we're looking at using MPLS with link coloring so that as we do start to get redundant paths available, we can ensure that non-ham-radio traffic stays off the ham-band links.
Matthew Kaufman
The most common place where I have encountered that would involve differing AUPs on different links.
For example, if one has a link which is built on an amateur radio layer 1, one cannot carry commercial, pornographic, encrypted, or certain other kinds of traffic on that link.
I believe Internet2 vs. public transit may also pose some such requirements.
Other situations I’ve seen involve data privacy concerns and/or security zone issues.
Common? Not in my experience.
Usually done with a combination of ACLs, Routing Policy, etc.
Owen
I don't think it is common, but I have a microwave network made up of a
combination of license-free links and amateur radio band links (where no
commercial traffic is permitted). For now the ham-band links are stubs, so
Are such Ham links actually of any real use, since encoded traffic
such as SSH/SSL
would be verboten, due to Part97 rules against transmitting any
message encoded
in order to obscure the message?
Also, with general network traffic..
If someone wants to request a Google search. There is no way of a router
knowing if the requestor is sending the packet for a commercial purpose or
for a non-pecuniary allowed usage, until TCP gets some new packet fields...
You can be visiting somepizzaplace.example.com, And it's non-commercial
allowed use, if you're ordering a pizza for personal consumption, But
those same packets are prohibited pecuniary use, if sending those packets to
order a pizza to share with a business client.
that's easy. But we're looking at using MPLS with link coloring so that as
Perhaps a browser plugin to add a 'Selection' dropdown for each Web Browser Tab
and have a RESTful API to send connection information from the client
to an Openflow controller for deciding which forwarding label to
push at ingress.
a fellow researcher wants
> to make the case that in some scenarios it is very important for a
> network operator to be able to specify that traffic should *not*
> traverse a certain switch/link/group of switches/group of links
> (that's true right?). Could you give some examples? Perhaps point
> me to relevant references?if so, why? security? congestion? other? but is it common? and, if
'Level3 Maintenance for Fiber path X on date Y'
where 'fiber path x' is one of your paths from A to B. Gracefully move
traffic (isis/ospf/rip/etc metric jackery), return traffic when the
crisis is past.
My experience has been with MPLS overlays.
Availability: During maintenance windows, moving high-value traffic away from potential outages while keeping the tunnels full of BE; manually manipulating MPLS tunnel affinities (though this could be automated fairly easily).
Congestion: Whenever traffic load spikes past a threshold; diffserv-aware TE to prevent certain classes of traffic from routing over links with limited bandwidth, handled automatically via auto-bw.
Preventing non-optimal tunnel paths. No transoceanic trombones, please; MPLS link affinities designed into the network.
-Scott
a fellow researcher wants
> to make the case that in some scenarios it is very important for a
> network operator to be able to specify that traffic should *not*
> traverse a certain switch/link/group of switches/group of links
> (that's true right?). Could you give some examples? Perhaps point
> me to relevant references?if so, why? security? congestion? other? but is it common? and, if
so, how do you do it?randy
In the wireless backhaul space I¹ve seen carriers that would prefer a
circuit to go down rather than take the long path on a ring between tower
and switching center. I assume they are concerned with some sort of
latency requirement. We used RSVP-TE with link coloring as the solution.
-Steve