Zte ZXR10 8900 Series User Manual page 110

ZXR10 8900 Series User Manual (IPv4 Routing Volume)
IP FRR Functions
IP FRR Working
Procedure
LDP FRR Overview
100 Confidential and Proprietary Information of ZTE CORPORATION
2. The time is to notice the unavailable event to router's controller
layer. It costs several milliseconds to 10-odd milliseconds.
3. The time is to take the corresponding responses to the un-
available node and link. The response implies that trigger and
floods the new link state updating packet, usually is several
milliseconds to 10-odd milliseconds.
4. The time is to notice the other nodes in network that the local
router link is unavailable. Each node costs several milliseconds
to a hundred seconds usually.
5. The time is to recalculate the trigger route. For IGP protocol
using Dijkstra arithmetic, the time is 10-odd milliseconds.
6. The time is to calculate the new routing information with line
card, and form the forwarding table. The cost time varies from
the different routing entries number, usually is several hundred
milliseconds.
7. The time is to record the new forwarding route entries into
hardware. It costs 10-odd milliseconds.
The traffic losing is occurred in the above mentioned steps, it can
be divided into two stages as follows.
1. Stage 1: The router can not discovery the unavailable link im-
mediately, and it still forwards the traffic to the unavailable
link.
2. Stage 2: The route has found the unavailable link, but the
network is in convergence process. The local forwarding table
is different with other router's, which causes "micro-loop" in
forwarding layer.
To decrease the traffic interruption period, a mechanism has to be
provided to realize the following requirements.
� Fast discovery of unavailable link.
� When the link is unavailable, fast provide a recovery path.
� Prevent forwarding loopback "micro-loop" occurs during the
further recovery process.
IP Fast Rerouter (IP FRR) technology is adopted to satisfy the
above requirements.
The working procedure of IP FRR is shown below.
1. Examine the failure fast. The common technologies include
BFD and physical signal examine.
2. Modify the forwarding plane. Switch the traffic into the recal-
culated backup path.
3. Perform route re-convergence.
4. After finishing the re-convergence, switch the route to the op-
timal path.
Obviously, the backup path is to fill the FRR gap up, which switches
the traffic to the backup next hop, to guarantee the traffic will not
be interrupted.
In MPLS area, the link or node failure causes the unavailable LSP
and the interrupt MPLS traffic. Similar with IP FRR,
mechanism to guarantee the traffic can be fast recovered in MPLS
area. Redirect MPLS traffic to the backup path that is specified by
IGP is doable, because MPLS traffic path is specified by IGP.
The working flow of LDP FRR is shown below.
1. The route protocol advertises the optimal and IP FRR backup
path to LDP, LDP records their corresponding labels into for-
LDP FRR is a