I am actually verifying if any new features have been released, that might allow tunnel setups and re-optimization to work based on actual available bandwidth, based on actual load, but assuming you are using OSPF or IS-IS CSPF during the set up and optimization, it will take all the information and for each link it will evaluate its eligibility based on the bandwidth that was reserved vs. the bandwidth that is reservable, so if you have an interface of 100 Mbps and you have 10% of it as RSVP reservable and then a percentage of it reserved, then you have only the remains of that available. In other words if you run regular IP traffic, it will not be subject to RSVP reservations and will not be used in calculations, so if you overload the link with it, the tunnels will still stay up. Just like you can overload the link by sending all kinds of traffic through the tunnels (that do not automatically adjust), but bandwidth that you reserve will be less than what you are sending out. And set up and holding priority is what you can use to keep the tunnels where they are, but unless bandwidth is reserved the mechanism does not know about its use.
Check out the NANOG web site for a bunch of presentations and the best book, which is the TE classic is Traffic Engineering with MPLS by Eric Osborne and I like MPLS Fundamentals by Luc De Gein. Reading some stuff on RSVP would be great too, but it will put anyone to sleep. Yan ________________________________ From: victor <[email protected]> To: [email protected] Cc: [email protected] Sent: Tuesday, September 8, 2009 12:42:55 PM Subject: [c-nsp] MPLS-TE and bandwidth reservation Hello I'm experimenting with MPLS-TE and have a question about reservation of the bandwidth on an interface. It's more or less clear that each tunnel can receive the necessary bandwidth and that it is consequently subtracted from the overall bandwidth configured for the interface. Therefore there can only be a finite number of TE-tunnels going through an interface. My question is: What will happen if I run a generic IP traffic over this interface in addition to MPLS-TE tunnels? How will this IP flow affect creation of new TE tunnels? What will have higher priority (or what will be dropped/rerouted first) when the interface bandwidth reaches it's max physical capacity? It seems that with an abundance of information about MPLS I can't find neither exact answer to these questions nor the algorithm behind this particular process so any references to white-papers, books, etc are greatly appreciated. Thank you in advance. Victor. --Using Opera's revolutionary e-mail client: http://www.opera.com/mail/ _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/ _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
