|
MPLS
Multicast Fast Reroute Multimedia Networks Group Dept. of Computer Science University of Virginia |
| General |
|
Overview People Sponsors |
| Software |
|
Overview MPLS Multicast Patch MulTreeLDP Testing Example |
| Documents |
|
MS Thesis |
| Contact us |
Overview
The MPLS Multicast Fast Reroute is an extension to MPLS that performs link failure recovery for multicast routing trees in MPLS networks. Repair of link failures in multicast communications can be performed within 50 ms (on average). The repair is done with precomputed backup paths for an existing MPLS multicast routing tree, thus anticipating the failure. When a link is cut, traffic is rerouted quickly from the initial path to the backup path. We provide an algorithm that computes backup paths which minimize the number of computers disconnected from the routing tree on a single link cut. Alternatively, backup paths can be computed manually.
What is MPLS ?
Multiprotocol Label Switching (MPLS) is a virtual-circuit packet switching technology created in 1997 for the transport of IP data. MPLS creates logical circuits that are called Label Switched Paths (LSPs). When an IP packet enters an MPLS network, it is first classified (according to its source, destination or other parameters) to determine the LSP to which this IP packet belongs . Then, MPLS adds a label to the packet that identifies the LSP. Each routers in the LSP uses this labelling technique to make a routing decision, that is, the routers decide the next router for the packet and changes the LSP. The last router of the LSP removes the label.
An advantage of MPLS is that it allows for good control over the traffic that flows over the network by influencing - manually or automatically - on the paths (LSPs) taken by packets. For example, by carefully setting LSPs, administrators can redirect traffic from congested links to links where more bandwidth is available. The set of these techniques is called Traffic Engineering; MPLS greatly facilitates deployment of such Traffic Engineering techniques.
For more information, please refer to the MPLS resource center, or the IETF MPLS charter.
What is Multicast?
Multicast refers to the network support for communication between a single sender and multiple receivers on a network. Group applications such as teleconferencing, file-sharing, video broadcast can greatly benefit from multicast support in a network.
Why Fast Recovery ?
In a communication network, one of the most common causes
of service disruption is link failure, caused for instance by a mere fiber
cut. After such a failure occurs, repairing a network so that service is
restored for end users can take a long time - a few seconds or minutes.
Implementation
We have implemented both the backup path choosing algorithm
and the rerouting technique described above, which we call MPLS Multicast
Fast Reroute in reference to its unicast counterpart dubbed MPLS Fast Reroute.
Below is a photograph of the testbed on which we carried all of our
experiments; it consists of 6 Linux PC-routers.
Our implementation consists of two separate parts. In the
first part, we
provide multicast extensions to already-existing MPLS-Linux. MPLS-Linux itself
is a project independent from ours, which comes as a patch for Linux kernel
2.4.10. Our software is distributed as a patch that adds multicast capability to
MPLS-Linux. The
second part of our software implements MPLS Multicast Fast reroute, the backup path
selection algorithm, and the "MulTreeLDP" protocol which sets multicast LSPs.
- Hardware of the Testbed System
- 6 x PII-450
- 6 x 128 MB RAM
- 6 x SCSI-2 hard drives
- 6 x 1 fast ethernet interface dedicated to administration tasks, connected to the University LAN
- 6 x 4 fast ethernet interfaces (Adaptec Quartec, aka Starfire), configurable on any (private) LAN
- Ability to make physical connections between any of the 24 private interfaces thanks to a patch panel
People
This work is supported in part by the National Science Foundation under grant number ANI-0085955. MPLS Multicast Fast Reroute is part of the Denali project for scalable communication services for the global network.
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation or any other sponsor.
