Skip Navigation
 Search | Directories | Reference Tools
UW Home > UWIN > Computing and Networking 

Spanning Tree Enhances Stability of UW Networks

Date: June 15, 2007

During the 2007 summer, in an effort to significantly reduce the impact of looped network ports on campus-wide network services, C&C will begin implementing the Spanning Tree Protocol (STP) on its layer 2 switch infrastructure.

What is a Looped Port?

A looped port occurs when two ports on a given LAN are connected to one another. The result of looping two ports on a given LAN is that all Broadcast and Multicast traffic on the LAN is repeated by the looped port, thus generating a 'traffic storm' that eventually consumes the entire bandwidth of the LAN.

Loops can be formed on a LAN in a number of different ways. The most common scenarios we have seen over the years are the following:

  1. A single cable connected between two 'live' Ethernet wallports
  2. A cable connected between two ports on a client owned hub/switch
  3. Two client/owned hub/switches, each with an independent uplink to the LAN, connected to each other

Higher Speed Means Greater Impact

As the University of Washington's data network continues to grow and faster data transfer rates are more widely available to users, the impact of looped port problems on the edge of the network can be much more significant than ever before.

Since the impact of these loop occurrences has become greater if not more frequent, Computing & Communications (C&C) has spent the better part of a year reviewing many technical options for reducing or completely eliminating the possibility of network loops and their corresponding impact to the campus network at large. After a significant amount of testing it was decided that implementation of STP throughout our layer 2 infrastructure would be the most economical and effective option with the least amount of impact to our clients.

About STP

The Spanning Tree Protocol (STP) while originally developed to allow for redundant architecture design at the layer 2 level has the added benefit of providing a fair amount of protection against Broadcast/Multicast storms caused by looped ports. The basic mechanics of STP revolve around STP specific management packets broadcast throughout the LAN from all STP enabled devices. These packets or Bridge Protocol Data Units (BPDUs) contain the critical information that allow the layer 2 switches to build a logical picture of the LAN topology. Thus when a loop is introduced to the topology, the switches are able to detect this loop and disable one of the two "redundant" ports, preventing the ensuing data storm.

How Will the STP Project Affect You?

C&C plans to enable STP on more than 3000 switches on it's layer 2 infrastructure over the coming months. For the most part this process will be invisible to our clients, aside from the significant reduction in widespread outages caused by looped ports. However, some switch models will require a brief service interruption to fully enable STP. In all cases C&C will issue standard 'Planned Events' messages notifying subnet contacts when this work will take place. Any concerns or questions can be directed as a response to these planned events, or directly to the Network Operations Center.

Many departments throughout the University manage some level of switch infrastructure beyond C&C's edge devices. In many cases STP on our switches will still be able to detect loops beyond our edge and take the appropriate actions to ensure these loops do not impact the LAN or the greater campus network. The only caveat being that the port blocking will happen at our edge and thus block the entire department managed switch which contains the loop beyond.

STP and Departmental Layer 2 Switches

Some departments support managed layer 2 switches that can be configured to participate in the LAN spanning-tree. At this point C&C neither discourages or advocates this participation, so it is up to the discretion of the department and their IT staff to decide if they want to extend the tree across their devices as well. If departments do choose to implement STP on their equipment C&C does have a number of requirements. Failure to comply with these requirements could result in either an intentional disruption of service to ensure network stability, or an unintentional disruption caused by incorrectly configured STP parameters.

Requirements for participating in C&C managed layer 2 spanning-tree:

  1. In order to comply with the capabilities of all our supported network devices, C&C is running standard STP, not RSTP or MSTP. Any clients wishing to participate should be running STP as well in order to ensure compatibility.
  2. To ensure network performance, stability and availability, it is critical that the C&C managed switches operate as root of the spanning tree. To ensure this, clients should not change the default settings of any of the STP parameters on their devices, except the aforementioned STP mode and possibly enabling 'portfast' if they wish. If you have questions or concerns about whether the default settings of your devices will comply or create problems, you can send you settings to customercare@cac.washington.edu with a request that they be reviewed by C&C Network Operation Center or Network Services.
  3. C&C is running Spanning Tree to help reduce the impact of loops to the network, not to allow engineering of looped devices for redundancy. Designing intentional loops for redundancy is more complicated and if done incorrectly can have a significant impact on the performance of the LAN. Any departments with a demonstrated need to design layer 2 redundant loops should submit a departmentally approved request with a detailed design to customercare@cac.washington.edu for review and approval.

For more information, send email to customercare@cac.washington.edu or call the C&C Network Operations Center at 206-221-6000.