Cisco Blogs

Cisco Blog > Ecolibrium

Steps towards energy efficient networking

As a follow up to my last blog, I’m happy to say that ATIS has taken the final step in publishing a groundbreaking standard for networking energy efficiency. For those who like the details: I’m referring to the newly published American National Standard ATIS-0600015.03.2009 — which most people know as Telecommunication Equipment Energy Efficiency Reporting (TEER). In the meantime, the International Telecommunications Union (ITU) Focus Group on climate completed a successful meeting in Hiroshima and is making excellent progress towards international standards for energy conservation in networks, and there’s movement in EPA to include networking products in their excellent Energy Star program. I’m proud to say that Cisco people are major contributors to these industry wide initiatives and I think these will be important milestones on the road to improved energy efficiency.Metrics are an important part of our drive towards energy savings, but we also need to look at where they fit into the bigger picture.Within Cisco we have looked at energy savings for ourselves as well as for networking users in general. We have found that a methodical approach helps to understand how and where efficiencies can be improved. There are many cases where a network can help you to save energy by eliminating the need to ship physical items (such as documents, CD’s or even people) when a virtual shipment is possible. I’ve saved many hours of air travel by taking part in TelePresence meetings or Webex calls. However, my expertise is not in these savings but in improving the efficiency of the network itself. The four steps to network energy efficiency that I recommend are:• Measure the current energy usage• Understand the productive network function• Analyze the network architecture and components• Compare solutions at points in the networkThe first two could be taken in either order; some of my colleagues would reverse them. I have chosen to start with measuring current energy usage as I am such a big fan of metrics and also because I believe that the measurement can also serve as a justification for the cost of efficiency improvements. It is important to measure as much as possible with as fine a granularity as is feasible. Energy budgeting is the same as any other type of budgeting — you have to start by seeing where it’s being spent right now.The second step seems like a no-brainer — except that networks have grown up and evolved and picked up a whole bunch of applications along the way. In most organizations there are business critical functions that are completely reliant on the network. Many times these will be diverse applications and in diverse parts of the business. It goes without saying that an attempt to improve network efficiency that perturbs the primary functions of the business will result in unpleasant words with senior management…Once you have identified where the energy is being spent and you understand what the network is doing for you then you are ready to look at the architecture and components that you have in place. When looking at network components it is important to remember that everything talks to everything else, so how the parts interact is crucial. This step requires the deepest cross-functional knowledge. You can look at how the network functions are distributed – core vs. edge; how discrete components compare with multi-function devices; what end-to-end functions could help your network performance or energy conservation (e.g. application partition or Cisco EnergyWise); etc. This analysis will show how you can evolve the network to a more efficient architecture and can also show up a few points in the network where optimization will be most effective.So the final step is where we introduce the device metrics. We believe that the methodology captured in the new ATIS standard is a good example of how metrics can be applied at this stage. It is important that the metrics do not undo the benefit of the previous three steps. In particular, the efficiency metric must evaluate how a device performs the real function at the specific point in the network (otherwise the work of the second and third steps is wasted). The evaluation of efficiency at a point in the network may indicate that an immediate upgrade would be a wise investment or may be incorporated into a normal upgrade cycle. Hard data on the real energy usage will help to quantify the benefits of the exercise. As part of the efficiency evaluation, it is important to take into account specific features that a device may have that allow it to save energy beyond the simple measured power usage. Such features may ease or automate power savings, increasing the probability that the savings will be seen in real life. Also, features that ease the load on other devices need to be taken into account as they would have been during the third step. Lastly, features that explicitly allow energy savings elsewhere in the network or attached devices are a major plus.The energy efficiency metrics take into account the performance of the device. In most case this is expressed in terms of the maximum useful bandwidth. The key term here is “useful.” When you look at a specific point in the network, you will see that the means of assessing this throughput will vary. For example, a core device will need to route between any ports and must be able to deal with a very large number of streams; an edge device will be largely limited by its upstream bandwidth; in some cases you should expect an edge device to be oversubscribed so that it can support high bandwidth bursts from its edge ports without wastefully low utilization of the uplinks.Once you have completed these steps, you should have some well supported conclusions. You can make improvements to the network energy efficiency while keeping your operations and finance colleagues happy. In many circumstances, being Green means eliminating waste and positively improves the business bottom line.Hugh Barrass, Technical Leader, Cisco

Comments Are Closed