Executive Viewpoint 2017 Prediction: Ciena – Get Ready for Tomorrow’s Self-driving Network
It’s well understood that ever–increasing cloud computing, mobility and video demand is having a significant impact on our global networks. Soon enough we’ll be looking at system overload due to increasing Internet of Things (IoT) traffic, emerging 4K, 8K video and virtual reality applications. In fact, according to the Ovum Network Traffic Forecast: 2015-20, traffic will grow 25 percent annually over the next five years, surpassing one million petabytes per year by 2020. In order to keep up with today’s web-scale demands, making sure the network works smarter, not harder, will be key.
Today we talk about optical transmission rates in terms of tens and hundreds of gigabits per second – 10G, 100G, 400G – and before we know it, terabits will dominate the conversation. This extra demand is putting the pressure on operators, who need to scale their networks to respond in real-time to unpredictable traffic demands. However, it isn’t realistic to simply add networking resources in the same way it has been done in the past – there are capacity and cost limitations that must be considered. It’s time to understand how the network’s optical layer – powered by advanced software – comes into play to meet web-scale demands.
Moving Past Traditional Network Designs
The traditional method for designing networks involves manually inputting parameters into a planning tool or spreadsheet, then adding optical margin (or extra performance capability) to the original design to account for system degradation or for unpredictable spikes in demand. Instead of using this inflexible and wasteful approach, the network of tomorrow will be smart enough to understand its existing level of performance and automatically adjust or self-optimize so that the needed capacity can be achieved at any moment. Just as self-driving cars bring the promise of less congestion and fewer delays, networks will become self-driving to ensure 24/7 connectivity with robust speed and a high quality of experience that customers now expect.
Advancements in coherent optical technology, including new tunable optics and advanced software capabilities built into the optical layer, will allow the network to become more agile and programmable. In most cases today, operators do not have real-time data from the network or the necessary analytic tools to make fast, informed decisions. This leads to inaccurate capacity assumptions, which results in sub-optimal use of network resources and reduced profitability (leaving revenue on the table). Soon, this will all change: The intelligent network is the promise of big data realized – network monitoring and information gathering in real time, combined with automated, intelligent decision making.
The Optical Layer and its Role in the Self-driving Network
There are two equally important ways the optical layer can be used to enable the self-driving network:
1. Deterministic Path Flows
It’s no longer sufficient to simply focus on adding capacity to a set of network links. Having abundant levels of capacity is one thing, but without having the insight into where the capacity should be directed, leads to waste and inefficiency. Adding intelligence to the optical layer makes it possible to understand how and when to shift capacity around. One route may only support an unprotected connection for basic services whereas an alternative route may support high availability for mission critical services. Understanding such options are important. For example, when the next Pokemon GO is introduced, the network must be able to adjust quickly so that users will not experience delays or lost service.
2. Re-shape the Level of Capacity in Real-Time
The ability to automatically shift capacity around without delay or human input is key to constituting a truly self-driving network. Now with the adoption of 5G, users are more apt to be able to access content whenever, wherever, requiring a network that gives us the bandwidth we need, when we need it. While a network operator can predict in advance the increased capacity demand for large, scheduled events, it’s the more organic upticks that are harder, or even impossible, to forecast.
The other part of maximizing resources is taking advantage of unused capacity to improve cost efficiency and service experience. Old systems were specified to operate with a certain fixed data rate, such as 100G, over a maximum distance, like 2,000km. However, links in a real network vary widely – they can be 95km or 372km or 1,125km. As a result, many of these older systems were over-designed for those links, wasting network resources. Now, with self-driving networks, intelligent systems can adapt the data rate for the appropriate distance. For example, when data only needs to travel 95km, the data rate can be tuned to 400G, increasing available bandwidth without changing equipment. Self-driving networks take advantage of the technical capacity-distance trade-off. Like a balloon, squeeze at one end (shorter distance) and the other end gets bigger (higher capacity). Even in a failure situation, intelligent decision-making can adjust the data rate on the fly so when the network is forced to reroute along a longer backup path, the network can still operate services with reduced link capacity – only degrading the service bandwidth without failing completely.
2017 will be the year that the foundation for tomorrow’s self-driving network will be laid, and the focus will shift toward not only adding capacity and increasing profit margins, but also improving the performance of the network via advanced software techniques. Ultimately, moving to an automated model for traffic management will ensure customers enjoy seamless connectivity. Capacity can be deployed where it is needed most so that no one misses out on the latest Netflix series at peak periods, while also ensuring that complaint calls remain low and satisfaction and loyalty remain high.