Print this article

Comment

01 December 2004

Proprietary rebounds faster

Ethernet moves to more plant floors as we speak.

The office automation with its old spanning tree technology is too clumsy for factory floor use, and there's much interest in the development of fault-tolerant or redundant industrial Ethernets.

Bennet Levine from Downers Grove, Ill., writes InTech with some fundamental information on the topic.

What is Ethernet network redundancy? Ethernet network redundancy is the ability of the network to survive a single cable failure in its switch-to-switch links. The network survives by providing alternate data path(s) when a cable fault occurs.

Why should I care about redundancy? Network redundancy is important to you if you have a system or process that is highly integrated and a failure in the communication links can result in disastrous consequences, such as production loss, poor quality, equipment damage, or danger to personnel.

What types of redundancy are available? The Institute of Electrical and Electronics Engi- neers (IEEE) has published two protocols that deal specifically with network redundancy: Spanning Tree Protocol (STP) (IEEE 802.1D) and Rapid Spanning Tree Protocol (RSTP) (IEEE 802.1w). Also available are many proprietary ring technologies and trunking schemes.

What is STP redundancy? STP allows networks to wire-up in almost any topology. STP normally provides network recovery times of thirty to sixty seconds.

What is RSTP redundancy? RSTP is an updated form of STP and is backward-compatible. This protocol provides a faster recovery time—generally, in one to two seconds.

What is network recovery time? Network recovery time is the time it takes to restore the network after a cable failure. The faster the recovery time the better.

Why should I care about the network recovery time? A few minutes lost in an office environment is merely annoying and inconvenient, but even a few seconds' interruption of an industrial communication network can result in thousands of dollars of lost production. To maximize system reliability, most proprietary ring networks self-heal in less than three hundred milliseconds (ms).

What is a ring network? A ring network is simply a bus with one extra link that connects the last switch to the first switch. The ring network requires that each switch supports a redundancy protocol. Otherwise, messages would travel around the network indefinitely. Ring protocols generally disable one link (the backup link) to stop messages from circulating the network. When a link in the ring fails, the backup link cuts in to restore the network.

What is trunking redundancy? Trunking allows switches to interconnect with multiple parallel cables. The more cables you add between switches, the more bandwidth you achieve—and with selected products, the more levels of redundancy you provide. With two cables between two switches, you have one level of redundancy. With three cables you have two levels of redundancy and so on.

How do STP/RSTP and proprietary rings compare? A number of vendors offer STP and RSTP. Generally, both protocols require over one second for recovery. To achieve the fastest recovery time with RSTP, you must connect the network as a mesh. A mesh requires at least three connections between each switch and neighboring switches. Ring networks provide faster recovery times than STP/RSTP, and to support the ring each switch uses fewer ports than do the switches in the mesh topology.

Nicholas Sheble (nsheble@isa.org) edits the Networking & Communications department. Bennet Levine (blevine@ccontrols.com) is a research and development manager at Contemporary Controls. He is a member of the Open Device Networking Association's Ethernet/IP Infrastructure Task Force. See his Web sites www.ctrlink.com/abc.htm, which presents basics of managed Ethernet switches, and www.rapidring.com, which addresses redundancy of networks.