T
he wide variety of features available in Ethernet and TCP/IP
(Transmission Control Protocol/Internet Protocol) are often
not used. This leads to a network that barely caters for the ap-
plication for which it is intended, let alone for any future upgrades
or expansions. Spending extra time and effort researching what is
available; and properly designing, planning and commissioning the
system, will lead to a strong communications network that provides
a high level of security and reliability, and ultimately to savings on
maintenance and downtime imposed by the communications network.
Safety and productivity will increase making it possible to cater for
future upgrades or expansions.
Redundancy
One of the most critical points that must be addressed when design-
ing a properly reliable network is redundancy. For a mission critical
network, any kind of communications failure can lead to shutdowns
of production or control systems for safety reasons. Preventing this
by applying proper redundancy to the network is essential.
When one refers to redundancy in terms of communications, the
reference is to link redundancy, the redundancy on which this article
focuses. As the redundancy used will depend largely on the topology
of the network (or vice-versa), these are two of the first details that
should be considered when designing a new network (or expanding
an existing one).
We need a redundancy protocol because the Ethernet protocol
does not function correctly when a logical communications loop is
introduced. This is due to Ethernet requiring broadcasts (among other
traffic types) to function. When a switch receives a broadcast, it will
forward that broadcast out of every one of its ports connected to the
same network - except the port on which it received the broadcast.
Any network device that receives this broadcast is obliged to inspect
the broadcast message; and even if the message is not relevant to
that device, it has to spend time and processing power to open the
packet and then discard it. If a loop is introduced to a network, these
broadcast packets will circle the network indefinitely, leading to an
eventual crash of the entire network. Redundancy protocols solve
this problem by logically disabling physically connected links, while
keeping them in a standby mode. If an active link fails, one of the
redundant links will be activated so that communications across the
network are not interrupted - at the same time eliminating any loops
that may form.
Figure 1: The four basic topologies in an Ethernet network.
Figure 1 shows the four basic topologies that can be found in an
Ethernet network (an actual physical network may consist of a mix-
ture of these topologies). As can be seen, a line - or a star - topology
provides no link redundancy, as the failure of a single link will lead to
communications being interrupted between segments of the network.
A ring topology provides a single level of redundancy, as a single
link can be lost without losing communications. However, a further
lost link will interrupt communications. A mesh topology provides
the highest level of redundancy, as the network can afford to lose
multiple links (depending on the overall level of meshing present on
the network) before communications become interrupted.
Different redundancy protocols work on different topologies.
Many proprietary protocols provide extremely quick recovery times
but will only work on a ring topology. The most commonly used open
redundancy protocol is RSTP (Rapid Spanning Tree Protocol), which
works over both mesh and ring topologies (any protocol that works
over a mesh will work over a ring without a problem). Another point
to consider when selecting a redundancy protocol is the possibility of
becoming vendor locked. A proprietary protocol may provide quicker
recovery times, but if you wish to upgrade or expand the network, you
may find that you are limited to buying that specific vendor’s products,
or alternatively, you may need to replace all networking hardware
on the network. It is also important to make sure the redundancy
protocol for which you have opted caters for the number of switches
in your network, as well as for any future expansion of the network.
Another way in which the effectiveness of a redundancy protocol
is measured is the recovery time of the network. This is generally
measured as the worst case recovery time, as failures at different
points in the network will take different times to recover, and will
Reliability and security
on
modern Ethernet networks
By T Craven, H3iSquared
Reliability and security are key to planning and designing a communications network for mission critical communications.
CONTROL SYSTEMS + AUTOMATION
Electricity+Control
July ‘14
4
