Network recovery

Layer 2 mechanisms to increase reliability

Spanning tree

IEEE 802.1D spanning tree is an Ethernet loop avoidance protocol. It allows network managers to connect redundant network links within their networks. Prior to the advent of spanning tree, loops within a switched Ethernet network would forward traffic around the loop forever, which saturated the network and prevented new traffic from getting through. Spanning tree selects one switch as a root and creates a loop-free topology connecting to the root. If loops are discovered, one switch blocks that port until its alternate path to the root is disrupted. Then the blocked port is brought back into service. There are several drawbacks to spanning tree:

By default, all switches have the same priority, which means that root bridge selection can be suboptimal in a network.

Spanning tree is slow to converge. It typically takes at least 50 seconds from link failure for a backup link to become active. As Layer 2 complexity increases, so does convergence time.

Although there are mechanisms for speeding up spanning tree, most are proprietary.

Traditional spanning tree is not VLAN aware. Thus, it will block links even if VLAN provisioning would have prevented a loop.

To solve these issues, the IEEE has recently introduced 802.1s and 802.1w enhancements. 802.1w introduces rapid spanning tree protocol (RSTP). RSTP uses active handshaking to speed up convergence times. 802.1s introduces multiple spanning trees (MST), which is a way of grouping different VLANs into different spanning tree instances. These features might not be present in data network switches yet, but look for them soon.

Link Aggregation Groups

Link Aggregation Groups (LAGs) are a mechanism for combining multiple real inter-switch links (typically four, Avaya products are configurable from two to eight) into one point-to-point virtual interswitch link. The advantage of this mechanism over spanning tree is that an organization can have the redundant links in if a failure occurs in one of the LAG links, the two switches will quickly discover it, and remove the failed link from the LAG., which reduces the convergence time to nearly instantaneous. Not all implementations interoperate, so care must be taken when the LAG connects switches from multiple vendors. Also, LAG links are a point-to-point technology. They cannot be used to connect a backup switch in case the primary fails. When available, this is a very good mechanism for improving the resiliency of LANs.

340 Avaya Application Solutions IP Telephony Deployment Guide

Page 340
Image 340
Avaya 555-245-600 manual Layer 2 mechanisms to increase reliability, Spanning tree, Link Aggregation Groups