JUNOS High Availability

Non-Stop Bridging

While most of the Juniper-deployed infrastructure on the Internet provides only routing functionality, more and more networks are deploying Juniper gear for switching purposes as well. Having a highly available switch is as important as the router. The bottom line is that networks provide only a medium for bits and bytes to move across. The failure of the forwarding card or control plane on the switch or the router has the same, unpleasant effect on the network. Therefore, it is necessary to provide some sort of system resilience for switches as well. JUNOS does it with the concept of Non-Stop Bridging (NSB), which, at the time of writing, is available only on the MX platforms. NSB is equivalent to NSR in the bridging world. The goal of NSB is to provide support for control plane failures while maintaining the forwarding state and keeping all neighbors blissfully in the dark about any failure. None of the Layer 2 control protocols should be able to detect the failure of the neighboring switch. The under-the-hood infrastructure of this concept is very similar to the NSR infrastructure.

Implementation Details and Configurations

NSB relies heavily on GRES for its forwarding state maintenance. As described in the previous section, kernel entries must be copied from the primary RE to the secondary RE. However, in this case, kernel entries contain Media Access Control (MAC) addresses and their respective next hops. Nevertheless, the replication process is the same. The ...