Energy Efficiency Made Simple Vol IV 2015

of the newer open standard redundancy protocols to be released. Either HSR, PRP or HSR (High-Availability Seamless Redundancy) and PRP (Parallel Redundancy Protocol) work by transmitting data along two different paths to its destination. This means that in the event of a failure on Path A, the data will arrive via Path B, without any further delay or requirements for retransmission of the data. In other words, bumpless redundancy does not require time to recover from a failure. These two redundancy protocols provide zero recovery time, which is great for mission critical, latency sensitive applications, however, the capital cost will be more than required for standard redundancy hardware. When it comes to cable redundancy, it is best to look at it from the perspective: How much redundancy can you afford not to have? Will a cable break lead to complete shutting down of the entire grid (or a section thereof) or is it simply going to mean a few people lose out on some non-critical monitoring data. Installing an extra cable and redundancy to prevent a countrywide blackout is definitely worthwhile; however, installing the same cable to ensure a single user has 24/7 access to personal emails is not. These are obviously extreme examples, but this is the decision process for any redundancy. Weighing the cost (whether money, time or effort) verses the potential losses if the redundancy is not in place will determine if it is worth it. There is potentially no upper limit to re- dundancy (although using every port on a switch to create a redundant connection to another switch is obviously not useful at all). Budgetary considerations will help determine the specific limit for your network. Whilst the best option could be to run redundant hardware at every point in the network, with redundant power supplies and multiple redundant uplinks between sections of the network, this can quickly deplete even the largest of budgets. Having little to no redundancy could mean incurring huge losses by the simple accident of a cable breaking. Once again, finding the balance is the biggest trick here, and should not be undertaken by anyone without a working knowledge of Ethernet and the redundancy options available. Hardware redundancy Cable redundancy is a topic that warrants special attention when planning and designing a network. In fact redundancy is one of the core decisions of the network as it can affect many other decisions. Cable redundancy is not the only redundancy available, even though it is the most commonly discussed. Another big redundancy point that should be considered is hardware redundancy options. Whilst cable redundancy protects against a communications link or cable breaking, hardware redundancy is more concerned with what happens if a piece One of the simplest of these, yet an overlooked or misused option, is power supply redundancy. Units with dual redundant power supplies, when installed properly, will prevent equipment from shutting down in the event of one power supply failing (and provide a level of load sharing between the two supplies, thus extending their lifetimes). It is important that this is implemented correctly. For instance, daisy chaining power from a single supply to power both redundant inputs defeats the purpose. Although this will prevent shutdown if only a single power input on the device fails (although in some cases even of hardware or a component fails. Power supply redundancy

PC A

Switch A

Switch B

1

2

4

1

2

Switch D

Switch C

PC B

2

1

Figure 2: Basic star topology – No redundant links available.

In Figure 2 it can be seen that we have solved the issue of the broad- cast storm and have changed the network backbone (connection between switches) from a ring to a line/star topology. However, what happens if we now lose the cable connecting switches B and C? Communications between PC A and PC B will no longer be possible at all, as there would be no communications path remaining between their network segments. This leads us to the reason redundancy was created and is used, especially on mission critical networks. While cable redundancy mechanisms work in different ways, their outcome is the same: they allow us to have physical loops on the network, yet they logically disable connections so as to break any communications loops on the network (such as in the image below, a redundancy mechanism has effectively ‘broken’ the link between A and D, even though the physical cable is still connected). In the event that another cable break leads to communication interruption, the mechanism will attempt to re-enable any redundant links held as back-up so that communications are not interrupted.

PC A

Switch A

Switch B

1

2

2

1

1

2

Switch D

Switch C

PC B

2

1

Figure 3: Basic ring topology with redundancy in place.

There are many redundancy protocols available, a number of which are proprietary to certain manufacturers. RSTP (Rapid Spanning Tree Protocol) is one of the most commonly used open standards and is supported by most hardware manufacturers. However, RSTP can take up to 30 seconds to recover in a worst case example, and so can be unsuitable for certain applications. In these cases one may need to look at either a proprietary redundancy protocol, or alternatively one

58

ENERGY EFFICIENCY MADE SIMPLE 2015