Technology Update /
Ring Spanning Tree brings resiliency
Carriers everywhere are deploying Ethernet in their metropolitan-area networks, and the fiber they use is being deployed in ring configurations. This has given birth to a demand for Ethernet resiliency that is designed for rings. And that is available now.
A simple and increasingly popular solution for rings is Rapid Ring Spanning Tree (RRST). An important enhancement to Spanning Tree Protocol, RRST is popular because it represents an Ethernet-802.3-compatible protection system, optimized for ring topologies.
RRST has gained attention as carriers in the U.S., Europe and Asia have deployed it through their metropolitan Ethernet networks.
Not to be confused with RPRRRST isn't the same as Resilient Packet Ring (RPR) protocol, although it shares many of the same goals.
RPR represents an effort to design a new Layer 2 protocol for highly resilient data transport.
RRST, on the other hand, works with "classic" 802.3 Ethernet - the strain found in enterprise and metropolitan networks. This means it can be implemented directly in most metropolitan Ethernet networks.
It is expected that RRST and RPR will be deployed in different parts of service providers' networks. Early versions of RPR are primarily deployed as regional transport solutions - that is, to connect carrier points of presence across an entire metropolitan area.
RPR has better resiliency than RRST, justifying its higher cost.
Meanwhile, RRST is better suited to being rooled out in access networks - networks that reach customer sites only within a city. In access networks, simplicity and low cost are the overriding concerns.
RRST takes advantage of several enhancements to the traditional Spanning Tree Protocol, known as 802.1D. It utilizes the recently standardized Rapid Reconfiguration Protocol, known as 802.1w, to make rapid decisions to change a port state from blocking to forwarding, and vice versa.
In addition, RRST works in a strict ring and therefore makes decisions for changing ports very quickly. The result is very predictable and fast (less than 500 msec) recovery times in most failure modes.
The protocol creates a separate resiliency instance for each ring in a service provider's network. This divides a larger problem (network resiliency) into multiple, smaller problems (ring resiliency).
Within each ring, RRST uses Spanning Tree's standard Bridge Protocol Data Units to establish a root bridge. Unlike random topologies with Spanning Tree, in which the root bridge almost never blocks traffic, in RRST the root is doing the blocking.
Because all RRST nodes are in a ring configuration, there is no need for complicated topology discovery and long waiting periods after a failure. The root bridge can immediately change the state of a previously blocking port, to forwarding.
It is also a useful tool for scaling Layer 2 networks. When service providers can deal with Ethernet resiliency one ring at a time, they can more easily build out large, multiple-ring networks while retaining the same predictable failover functionality.
Finally, ring spanning trees, while timely, are not the only important enhancement to spanning trees. Besides the well-known rapid spanning trees, per-virtual LAN multiple-VLAN spanning trees can be very useful for certain network architectures.
Wu is technical marketing director at Riverstone Networks. He can be reached at firstname.lastname@example.org.