The protection and recovery mechanism of MPLS TE: Principle Edition

In this document, there are several things that are involved:

Fast Reroute Frr--fast Re-route

Link protection

Node protection

Path protection

Some other advanced protection mechanisms.

Before we begin, we need to understand why protection and recovery are needed.

Some of the networks sometimes fail, and these failures involve a wide range of lines, fiber loose/broken to router crashes.

In IGP, if network turbulence is encountered, OSPF or other routing protocols recalculate the topology, then red in the zone, and finally synchronize all the databases in the zone.

But MPLS te, which has the ability to keep traffic out of the shortest path driven by IGP, provides fast rerouting FRR or MPLS te protection (simply Mpls te Protection).

In large networks, the IGP protocol requires a considerable number of seconds to a few 10 seconds to converge and converge. In the entire network stability before, for the customer's data business is in the vacuum phase, if the network oscillation flaping, the occurrence of 5-20 seconds Packet packet loss is very normal.

And IGP, a link-state protocol, once the link fails and then resumes, SPF recalculates all the topologies in the zone and synchronizes it to each router, during which time the synchronization topology makes the device a lot worse, and the CPU utilization runs to 80%. The topology stabilizes to normal forwarding, This is only IGP, and this problem is even more serious in Mpls te, because after IGP is stable, LDP or RSVP can work, build up and then distribute tags to create FEC, and so on, it takes more time than the IGP environment.

This is the knowledge module that this document needs to introduce, how to effectively protect the data service in the MPLS TE traffic engineering fast and effective uninterrupted forwarding.

Here or a lot of people have the same problem, if the path in the network more than one to the right side, then this is not a protection?

A link fails, switching to another link.

However, this description is in line with the requirements of the telecom operation level of switching, non-stop switching.

Here are some of the main types of protection you'll start with:

Path protection, also known as End to end protection (End-to-end protection)

Link protection

Node protection.

The first thing to introduce here is path protection.

Cisco does not support this type of protection in the data (based on MPLS TE traffic engineering).

But the book was a few years old. I found the relevant documentation on the Cisco website and Cisco already supports path protection.

Http://www.cisco.com/en/US/docs/switches/datacenter/sw/5_x/nx-os/mpls/configuration/guide/mp_te_path_prot.html

The document is written in detail. I looked at it and found that it was a bit like an explicit path, which is the path protection.

The only difference is that the backup path serves only one primary path.

In essence, path protection is achieved by building an additional LSP in parallel with the existing LSP. Backup LSP Except in the case of failure, is not to carry traffic, so in fact, this backup is 1:1 ways, in your network, how many hops will need to backup LSP, the general network loss efficiency is far below 50%, so the path protection scheme of the expansion is relatively small, and the price is too high.

The following is an example of path protection principles: