Huawei S6700 Series Configuration Manual page 108

S6700 Series Ethernet Switches
Configuration Guide - IP Routing
After being enabled with OSPF Graceful Restart (GR), a switch can ensure continuous packet
forwarding if it restarts just for abnormities. In such a case, route flapping is avoided during the
short restart of the switch.
When a switch restarts OSPF, the GR Restarter does not age the forwarding information. At the
same time, the GR Helper keeps the topology information or routes obtained from the GR
Restarter for a period. This ensures that traffic forwarding is not interrupted when protocol restart
occurs.
OSPF VPN Multi-instance
OSPF supports multi-instance, which can run between PEs and CEs in VPN networks.
In BGP MPLS VPN, many sites of one VPN can use OSPF as the internal routing protocol. The
sites, however, are handled as being from different ASs. In this way, the OSPF routes learned
on one site are transmitted as external routes to another site. This causes a heavy OSPF traffic
and some avoidable network management problems.
In the S6700 implementation, you can configure domain IDs on a PE to differentiate the VPNs
where different sites reside. Different sites in one VPN consider each other as if they were
connected directly. Thus, PEs exchange OSPF routing information as if they were directly
connected through a leased line. This improves network management and enhances the validity
of the OSPF application.
BFD for OSPF
By default, in broadcast networks, the interval for OSPF to send Hello packets is 10 seconds; in
NBMA networks, the interval for sending Hello packets is 30 seconds, and the period for
advertising that the neighbor is Down is four times the interval for sending Hello packets. If the
switch does not receive the Hello packet from the neighbor before the neighboring switch
becomes invalid, it deletes the neighbor. That is, the switch detects the neighbor faults in seconds.
This leads to the loss of a large number of packets in a high-speed network.
To solve the preceding problem in the current detection mechanism, Bidirectional Forwarding
Detection (BFD) is developed. BFD can implement detection at the millisecond level. Instead
of replacing the Hello mechanism of OSPF, BFD works with OSPF to fast detect the adjacency
fault. BFD is used to notify OSPF of recalculating routes. This can correctly guide the packet
forwarding.
Routing Management (RM) module exchanges routing information with the BFD module.
Through RM, OSPF notifies BFD of dynamically setting up or deleting BFD sessions. The Event
message of BFD is delivered to OSPF through RM.
The process of establishing and deleting a BFD session is as follows:
l
l
Issue 01 (2012-03-15)
NOTE
Unless otherwise specified, "protocol restart" in this document refers to restarting OSPF in GR mode.
Process of establishing a BFD session: If BFD feature is globally configured, BFD is
enabled on an interface or a process, and the status of the OSPF neighbor is Full, OSPF
uses RM to notify the BFD module of establishing the BFD session and negotiate related
parameters of BFD.
Process of deleting a BFD session: When BFD detects a link fault, BFD generates a Down
event and notifies the upper protocol of the fault through RM. OSPF then responds to the
event and immediately deletes the adjacency relationship on the link. At this time, the status
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
4 OSPF Configuration
89