Control.com - Nerds in Control

What's differ between Fault tolerent ethernet and ethernet? What's the concept of Fault tolerent ethernet?

Can anybody help some documents?

Thanks in advance.

While there may be several answers to this innocent question, I will begin.

Fault-tolerance is the ability of the network to continue operation when any one
cable or component fails. One way to do this is with a complete second (or
redundant) network - cables, switches, interface cards, etc. This is expensive,
and requires that the software reject duplicate messages received, one from each
network when they are all working. Foundation Fieldbus HSE works in this way.

A less expensive way is to split the transmit and receive pairs of 10/100BaseT
twisted pair on Cat 5/5e/6 cables. From the fault-tolerant network switch port,
the transmit pair is wired to the first interface card's receive pair
termination. From that first interface card, the transmit pair is not wired to
the switch, but is wired to the receive pair termination of the next interface
card. That card's transmit pair is wired to the receive port of the next
interface card. And so on though all interface cards. The last interface card's
transmit pair is wired to the receive port on the switch. This forms a ring
network. If the interface cards support a second Ethernet port, it can be wired
in the inverse allowing any one message to flow around the dual ring in
counter-rotation. This requires the interface card logic to store and forward
all messages not addressed to it's own MAC address, similar to the token ring
logic. The switch logic must also know to send messages addressed to a ring
member to both ports by which the MAC address can be reached. Any cable cut or
device failure will not cause a message to be missed = fault-tolerance. This
form of network first appeared on motion control systems from Jetter AG, and has
been commercialized by Hirschmann and GarrettCom, among others.

Dick Caro
===========================================
Richard H. Caro, Certified Automation Professional, CEO, CMC Associates,
2 Beth Circle, Acton, MA 01720
Tel: +1.978.635.9449 Mobile: +.978.764.4728
Fax: +1.978.246.1270
E-mail: RCaro@CMC.us
Subscribe to the CMC Wireless Report at http://www.CMC.us
Web: http://www.CMC.us

Just adding a little to what Dick Caro already mentioned:

In the process control world high availability is important because downtime is very costly. Usually the control systems "DCS" have redundant controller cards, each with redundant Ethernet ports, to ensure that all main and backup controller cards can communicate with each other even if a port fails.

I think this kind of "port redundancy" is not required in other industries. Process control is tougher.

Cheers,
Jonas

Fault tolerant Ethernet means the network will not fail because of a single component/cable failure. Though according to Honeywell this is a unique name for one of their control networks.

Fault tolerant networks are built using redundancy of Ethernet connections and network infrastructure. In the case of a server or workstation, dual Ethernet connections are used. In the case of network infrastructure (switches, etc.) multiple switches are employed with multiple network connection, in what is generally referred to as a "fabric".

Redundant network connections pose additional requirements on the infrastructure, since the switches need to direct traffic based on changes in network topology. Also the switches need to prevent "loops" where network packets can be trapped, thus causing escalating network traffic. This means that managed switches must be used where fault tolerance is required.

The most common technique employed by managed switches are Spanning Tree Protocol (STP) and Rapid Spanning Tree Protocol (RSTP), though convergence after an component/cable failure can take minutes for the former and several seconds for the latter. So many vendors have implemented special techniques for much faster recovery, which in some cases can be as little as 10ms.

Most use a "ring" configuration, where all of the switches are connected to one another in a recirculation fashion. One link between two switches is assigned "backup" status. During operation a token is passed between the switches to check for link health. In the event that any switch fails or a cable is broken/disconnected, the backup link is activated to maintain network availability.

Often additional resiliency is achieved through component redundancy as well. Many of the industrial switch manufacturers feature redundant power inputs. So in the event that power is inadvertently disconnected or a power supply fails, the switch remains functional.

Paul...