In recent years, Ethernet technology has grown in leaps and bounds and has become a networking technology of choice for running automation networks in industrial, utility, military and ITS environments. However, as the technology progresses and becomes more complex, so does network planning, configuration and maintenance need to improve to meet that complexity. Gone are the days of simple point-to-point serial communications running automation, where loss of communication is relatively simple to diagnose and repair.
Many Ethernet networks currently in place at these types of sites were not planned for correctly, especially when considering future expansion and the possible loads that would be placed on the network. This means that although the networks would work well when they were first installed, once Ethernet took hold as the network of choice, the networks were expanded to cover more and more of the site. This leads to network slowdowns end device communication failure due to untimely responses from master devices, and even complete network crashes, all due to the larger amounts of uncontrolled data travelling on the network.
For this reason, proper planning of a network is an extremely important part of the process of moving to Ethernet. The most obvious planning is to take into account the number of devices that will be on the network, as well as an idea of the traffic amounts these devices will impose on the network. However, at the same time care needs to be taken to make sure that the network will also be able to handle future expansion. Provided there was correct planning on the IP subnet structure on the network (The number of IP addresses available within each subnet to assign to devices) it is a relatively easy task to add on an extra switch at the edge of a network, however it is not such a simple task to upgrade the main backbone of a network. For this reason, planning and forethought is highly important to avoid future downtime due to required physical network upgrade.
However, planning for the physical network and the required bandwidth is only the first step. Even the most meticulously planned network may not operate in accordance with the user’s requirements if not configured correctly.
Due to the rapid expansion and evolution of Ethernet, it can often seem daunting to plan and configure features such as VLANs, routing, multicast control and so on. For this reason networks are often configured as a single flat network, with no multicast/broadcast control, routing or VLANs to try and separate and cut down on unwanted data reaching end devices. Although this may work for very small scale, non-mission critical networks, for a larger or more traffic intensive network, or a mission critical network where responses need to be prompt and reliable, a flat network will not suffice.
There are several protocols built into Ethernet that can help with controlling the data on the network, and will largely improve the reliability and response times. One of the most important of these for any large, data intensive, mission critical network is VLANs (virtual local area networks). The use of VLANs, defined in IEEE Standard 802.3q, allows the network to be logically divided into sub networks, based on which device groups they need to communicate with regularly. This means that devices in one VLAN will not be able to communicate directly with devices in another VLAN. On a flat network, whenever a device sends out a broadcast message (a broadcast message is one that will travel to every device in the same subnet, and each device that receives that broadcast is obligated to inspect the packet to see if the device must respond), every other device on the network receives it. This means that PLCs will be receiving broadcasts from devices such as cameras, historian servers and any other network device sending broadcasts. The end devices will be devoting substantial amount of their time and processing power to inspecting packets that they have no need to see.
Using VLANs you can logically separate the network over the same physical cable, meaning that unwanted traffic in each VLAN, or group of devices, is eliminated. In the event that some devices in one VLAN do need to communicate with other devices in another VLAN, an Ethernet router can be introduced which will allow data to be routed between VLANs. For VLANs, pre-planning is essential, as too few VLANs and the broadcast traffic will still be travelling to devices it does not need to, but too many VLANs and routing between them can put unnecessary strain on the router in the network which could also cause problems.
Ethernet networks are constantly evolving and expanding. For this reason even the most carefully planned and configured network still requires regular maintenance. Standard maintenance that is done on any electronic device is obviously a must, such as checking that devices are still getting correct power, are not overheating etc. For Ethernet networks, however, there is more logical based maintenance as well that is recommended in order to keep the network in a stable, reliable state (such as checking switches and routers on the network for any errors).
For instance a standard practice should be to check the log files and alarm lists of the Ethernet switches on the network, especially those making up the Ethernet backbone. This should be done on a regular basis, such as once a month or even every two weeks. This can prevent small problems, such as a failed redundant power supply, from become a much larger problem (Such as if the second power supply fails and brings the switch down altogether). Logs and alarm lists should also be backed up and then cleared once they have been checked and any concerns addressed, as this will greatly aid with future troubleshooting.
A network management package can greatly aid in this practice, as it can be set to automatically download log files on a regular basis, as well as backing them up and then clearing them off the devices themselves. Some can even be configured to check the firmware of the devices and upgrade them if necessary. If you are using a network management solution, you can even go a step further. By configuring SNMP (Simple Network Management Protocol) on the devices on the network you can allow the device to notify the network management station when they have an error, which in turn can be configured to notify the network administrator on duty. Again this will allow problems to be identified and corrected in advance, rather than waiting for the problem to escalate.
Another maintenance step that can be a great preventative of network issues and crashes is by having the network regularly audited by a team of Ethernet network support specialists. Using hardware and software tools, along with specialised skills, a team such as this can analyse the traffic on the network at various times, and using this can identify problem devices and incorrect configurations before they cause critical network failures.
In today’s age, with Ethernet networks being used for mission-critical networks for everything, from mining to production to substation automation, network engineers and administrators cannot afford any network failures. For this reason, correct planning and configuration, as well as regular preventative maintenance are essential to having the peace of mind that a stable, reliable communications network can bring to the end user.
|Tel:||+27 11 454 6025|
|Fax:||+27 11 454 6024|
|Articles:||More information and articles about H3iSquared|
© Technews Publishing (Pty) Ltd | All Rights Reserved