Dual homing has long offered the potential to add redundancy to plant floor automation systems for reduced downtime – but only at a cost that put it beyond all but the most critical installations. Now, though, an innovative new technology is bringing the benefits of dual homing to mainstream control applications without the associated costs. David Moss of GarrettCom Europe explains.
Effective redundancy strategies have always been of prime importance in critical applications where a fault or failure could result in catastrophe. But today, even in more standard industrial applications, the simple financial costs associated with unexpected downtime are leading increasing numbers of manufacturers to look at introducing redundant systems to deliver the high availability needed to ensure profitability.
Designing and implementing high availability industrial networks has traditionally represented something of a challenge. The many popular serial fieldbuses do not typically offer fault-tolerant options, making the choices for redundancy on the plant floor limited, expensive and complicated. The move to Ethernet as the backbone for industrial automation networks has seen the increasing use of ring and mesh topologies for connectivity. These topologies ensure that a break in the media doesn’t bring down the whole network, but a fault in an individual switch can still cause problems. True, the individual switch can be bypassed by network traffic, but any devices and nodes connected to that faulty switch are effectively isolated.
Finding practical ways to provide for recovery from faults in nodes connected to critical devices has remained difficult. Either, such functionality is not addressed by the control product vendors, or it is addressed in only a limited and expensive range of products, such that creating an effective redundant network becomes a complex trade-off between up-time, cost and control device selection. In short, true redundancy has often been prohibitively expensive for applications deemed as essentially non-critical.
One option for achieving this level of redundancy comes in the form of dual homing technology. Dual homing adds reliability – and hence boosts uptime – by allowing a device to be connected to the Ethernet network by way of two independent connection points. One is the operating connection, whilst the other is the back-up connection that is activated in the event of a failure in the operating connection. The fault in the main connection path can be quickly identified and repaired whilst traffic on the network is maintained on the back-up connection, keeping the production line up and running.
In a typical industrial Ethernet network, various sensors, controllers and devices might be connected to Ethernet switches, and the various switches linked in a ring, dual ring or mesh topology. Dual homing technology brings a simple means of adding redundancy to all of these network structures.
A simple ring structure, for example, offers some ability to recover from a media break or the failure of a switch, but a failure of a switch in the ring takes down all the nodes connected to that switch. But dual homing allows a critical node to be connected to two switches in the ring, thus introducing redundancy and so greatly increasing overall reliability. And as well as protecting against network faults, dual homing in a ring also provides connection flexibility during maintenance work, allowing a switch in the ring to be taken temporarily out of service without interrupting uptime.
Some networks use two rings for improved redundancy, but dual homing technology can further enhance these topologies. Where a single Ethernet ring can recover from only one fault at a time, with two rings there are two paths, providing the potential to recover from more than one fault at any given time. Even so, whilst the rings are linked, any device or controller will generally only be connected to a single switch, so again a failure in that switch could potentially take down critical nodes. Dual homing, however, allows a critical node to be connected to two Ethernet switches – one on each ring.
Mesh structures offer the best inherent fault recovery capabilities, with multiple interconnects offering the potential to recover from multiple faults. Mesh complexity is almost without limit to allow the highest levels of reliability to be conceived, but of course the cost of equipment and cabling to construct the mesh rises as the complexity increases. And, as with rings, any critical devices connected to an individual switch will still be taken down by a failure in that switch. Dual homing technology allows any given device to be connected to two switches in the mesh, building in redundancy.
Low cost dual homing technology
Whilst dual homing has long provided a conceptually simple means of achieving high levels of redundancy, actually implementing such dual homing within an Ethernet network has traditionally relied upon having the dual homing functionality within the critical node itself – a PLC, industrial controller, sensor, etc. But very few manufacturers offer PLCs and devices with dual connectivity built in, and the rare dual-NIC devices available (typically computers and some SCADA devices) require complex software to manage the NIC-switching logic. All have been prohibitively expensive for general industrial applications. Whilst there may be many documented examples within industry of unplanned interruptions costing sums that extend into millions of pounds per minute, the risk of such interruptions has to be weighed against the cost of installations to guarantee against them.
GarrettCom, however, has overcome the cost and product availability problems associated with dual homing with the launch of a low cost, Ethernet switch that incorporates a new dual homing technology.
With the ESD42 dual homing Ethernet switch, the dual homing function is moved from the field node into the switch itself. Connecting to any Ethernet enabled controller or device on the plant floor, the compact ESD42 dedicates two peer ports to provide two network access points to protect against any single point of failure on the network. Detection of a media connection fault or an upstream switch failure simply causes the ESD42 to transfer upstream network activity to the other port – typically within 300 milliseconds.
The ESD42 is itself an extremely cost-effective device, but it also massively reduces the cost of implementing dual homing by removing the need to buy dual-homing enabled PLCs, industrial controllers, sensors, etc. Redundancy can be provided for any Ethernet enabled device: it is simply plugged into one of the local-device ports of an ESD42 switch, while the two dual homing ports on the ESD42 are connected upstream. And no dual homing controller is required to detect network failures and implement the connection transfer: instead, GarretCom’s patent-pending technology will detect Link Loss on the operating port and automatically switch all data transfers to and from the connected devices to the standby port, which then becomes the operating port.
There are no configuration variables for the user to worry about – the ESD42 provides a plug and play solution. On effecting a dual homing switch-over, the ESD42 automatically changes the state of its internal address buffers and notifies upstream devices of the connection change of the attached nodes to the new operating port.
In a standard automation network built on Ethernet, any given switch in the network is a potential point of failure. If that switch is connected to a number of control devices as well as to other switches in the network, then it is likely to be an inherently complex, multi-port Ethernet switch. As such, its risk of failure at any given time will be a risk appropriate to the level of complexity of the device. When an ESD42 dual homing switch is added to the network, then it immediately becomes the potential single point of failure instead of the complex multi-port switch. This brings a far greater level of robustness to the network because the ESD42 is an inherently simpler and more reliable device, with a standard MTBF of 30 years!
Further, because it takes advantage of Ethernet standards, the dual homing redundancy features of the ESD42 work with any brand of Ethernet switch upstream. And GarrettCom has tested the ESD42 with different redundancy software systems, including its own standards-based S-Ring Redundancy Manager software and Link-Loss-Learn (LLL) technology, as well as with industry standard IEEE 803.1x Rapid Spanning Tree Protocol (RSTP) redundancy software. When tested with a variety of brands of Ethernet switches running standard RSTP software, the typical redundant network fault recovery time was less than one second.
The ESD42 dual homing Ethernet switch fills a void in the network market by providing designers of redundant industrial networks with a simple, dependable and cost-effective redundancy tool. Paired with the ESD42, almost any industrial control device can benefit from redundant operation to create an overall installation that is inherently more robust without any added complexity.
The trade-off between up-time, cost and control device selection need no longer be an issue, allowing developers of even the most cost-sensitive automation systems to take advantage of network redundancy and reap the benefits vastly improved productivity.
|
