Sponsoredby:
Cisco
The dawn of collaborative
Challenges to the evolution of the network Service provider (SP) networks are undergoing major changes. Traffic continues to grow at an exponential rate – around 30-50% per year globally and much faster in some cases. At the same time, a growing percent of the direct and indirect revenues from the services are going to “over the top” (OTT) service providers, such as Google and Netflix, leaving SPs with almost flat revenues. This trend strains the business model of SPs, as the gap between the cost of the network and the revenues gained shrinks. Continuous innovation in modulation formats has previously helped control costs, by putting more information into a GHz of spectrum on the fibre: from 10Gbps to 100Gbps in a 50GHz spectrum slice, or an increase of approximately 10x in spectral efficiencywithin10 years. However, there is mounting evidence that this will become harder as we approach Shannon’s (non-linear) limit. Today the upper bound is around 200Gbps per 50GHz using 16QAM – and in a couple of years, as DSP processing power increases – we will achieve 400Gbps in such a spectrum slice. However this seems to be the end of this approach: higher order modulation formats will have a very low reach. As a result,
the industry is turning to parallel solutions, such as superchannels (parallel, tightly spaced channels), or SDM (parallel fibre cores). But both techniques are not expected to provide significant cost reductions. Raw bandwidth growth is just one of several trends that may be as challenging to the evolution of the network: Consumer traffic is now much larger than business traffic – skewing the required technologies towards more dynamic IP based technologies. The number of main bandwidth sources of this traffic is becoming much smaller. For example, in the US, Netflix traffic represents a third of the overall peak traffic. This implies larger traffic fluctuations due to failures of peering points The emerging cloud computing paradigm will make it easy to mobilize an application from one server to another based on power savings considerations, proximity to the users and commercial considerations, further increasing traffic dynamism. The Internet of Everything - which will turn billions of devices into active users of the internet - will have an unknown, yet dramatic impact on the network. The only clear conclusion that can be drawn is that it will be increasingly hard to predict traffic behaviour. This means that the
network planner will have to plan for the unknown, but how does one do this without significantly over-provisioning the network and further straining the SP business model? Addressing the challenges in the optical layer The aforementioned changes in the network imply that the optical layer will have to be streamlined, flexible, and reconfigurable: Streamlined: the increased pressure on SP margins implies that the future network must be as efficient as possible, and this implies removing as many network layers and the interfaces between layers as possible –in fact most core networks are already evolving to two layers model: a transport layer and a service layer. Flexible: the lack of ability to forecast how traffic will evolve implies that the network will have to be as flexible as possible in providing the right amount of capacity where it is needed using the most effective modulation format. Reconfigurable: since traffic patternswillchangemorefrequently, the network will have to support graceful release, redeployment, and reoptimization of resources. Without these capabilities, resource will sit idle and the cost of the network will grow well beyond the required cost.
By Ori Gerstel
The best architecture is a hybrid one, with
both distributed and centralized control elements
Required bandwidth Unused wavelength Used wavelength
Optical layer
(a) non-agile solution: 7 wavelegths =Sum(Max{AI,BI })
(b) agile solution: 5 wavelegths =Max{Sum(Ai),Sum(BI)}
(a) Normal state
Scenario I:
Core router
Transponder
100 100
300
Scenario II:
(b) Optical failure
200 200
100
Figure 1 - How network agility reduces cost
Figure 2 - Multi-layer restoration
18 | Optical Connections 2013 | www.opticalconnectionsnews.com |
Made with FlippingBook - Online catalogs