The volume of traffic traversing networks today is rising faster than revenues, sending operators’ revenue per bit spiraling downward. The really bad news, however, is that it doesn’t look like the situation is going to improve anytime soon.
Analysis by research firm Innovation Observatory projects that the effective revenue per megabyte of data will fall by as much as 60% in the coming years, as illustrated in the figure below.
Given this, operators find themselves in an environment where they must, in order to remain profitable, find ways to push down the cost of delivering each bit of data while at the same time squeeze out increased efficiencies from their networks.
Dealing with the demand
Operators are attempting to cope with the increased traffic volume in a variety of ways.
Some are adding new routers to their core networks. This can be a costly proposition, as routers are among the most expensive network equipment. Plus, they take up a lot of space and consume a great deal of power, adding to the cost of ownership.
Others are adding new high-rate 40GbE or 100GbE interfaces to the routers in order to improve the system ‘fan out,’ or capacity that can be transmitted. This is not always cost efficient, as with this approach operators may be paying up front for capacity they don't initially need.
Fortunately, a more cost-effective option is emerging for reducing core network costs: router traffic off-load (also known as router bypass). Router off-load is a strategy that allows operators to add capacity without adding more routers. It involves identifying traffic that passes through routers on its way to other destinations and off-loading the task of transiting that traffic to more cost-efficient network layers.
Router off-load explained
The router off-load approach employs the principle of providing transport in the most cost-effective layer.
In a typical core network configuration, all traffic that flows through a PoP (point of presence) is processed by a router – whether it is intended for that PoP or is just passing through to another PoP (‘transit’ traffic). This is inefficient because it utilizes the router, an expensive network element, to process traffic unnecessarily.
With router off-load, the traffic that does not need to be processed by a core router - the transit traffic - simply bypasses the router at a lower level, such as Layer 0 or Layer 1. This can be achieved in a couple of ways. Traffic that is eligible to bypass a router can be switched using OTN (optical transport network) cross-connects at Layer 1, or it can be switched fully at the optical layer using ROADMs (reconfigurable optical add-drop multiplexers).
As traffic is terminated anyway at Layer 0 and Layer 1, there is little to no additional cost associated with router off-load. Since the nature of IP traffic is dynamic, some level of connectivity between the routers is required to maintain network flexibility. This connectivity is defined based on the network traffic patterns and topology.
Operators who are seeking to avoid investment in the additional router capacity required to manage the volume of traffic that is anticipated in the future may find router off-load to be a viable alternative.
Determining potential cost savings
Because no two networks are alike, there is no hard and fast rule that can be applied to determine the degree of CAPEX and OPEX savings that is possible with a router off-load strategy. Each network will have a different optimal configuration based on its equipment, topology and traffic mix, and as such, the economics will vary for each operator.
Many factors must be taken into consideration when determining the potential value of router off-load, including:
Network starting point: The benefit of a router off-load approach will depend on the infrastructure that is already in place. Do routers extend from the core to the metro? Are Carrier Ethernet switches or packet-optical transport systems in place? Or is it made up of legacy SONET/SDH gear and ADMs? The investment case will vary if new optical equipment has to be deployed.
Projected traffic growth and patterns: The current network load and projected traffic growth must be taken into consideration. The amount of transit traffic on routers can range anywhere from 30% to 80%.Efficiency gains will be greater in heavily loaded networks with strong transit traffic growth projections.
Lack of statistical multiplexing: Circuit-switched networks cannot perform statistical multiplexing of core traffic, so the impact of the loss of this function should be weighed.
Architectural reconfiguration: In some networks, implementing a router off-load strategy to cope with future transit traffic growth can be as straightforward as simply not using routers. However, the optical infrastructure must be current and have control plane capabilities. Analysis has shown that the capex savings from eliminating just three routers can be up to 60% (assuming six extra router ports are needed to handle a growth of transit traffic equivalent to 10 Gbps over time).
Cost considerations: Other possible costs that need to be taken into account include necessary alterations to the WDM infrastructure, integrated management and control plane solutions and the mix of protection and restoration strategies.
Potential added savings: Router off-load can provide additional savings in the form of reduced power consumption, space requirements and operational costs. Data from operators suggests that the power consumption of optical switches is 10-20% of that of routers for the same transport capacity. Optical switches are much smaller than routers, so the need to invest in more floor space for additional routers as traffic grows is avoided. And OTN and optical layer provisioning can be done automatically, saving time.
Money saver or latest flavor?
Is router off-load a passing fad, or does it hold real potential for improving operators’ bottom lines?
It often depends on whom you ask. Academics and technologists alike have analysed the potential of router off-load using a broad range of models, and they have arrived at differing conclusions. A considerable body of literature has been produced either promoting or disputing the potential of off-load strategies.
What is clear is that there is no ‘one size fits all’ solution when it comes to networking. No single solution will suit all operators.
But what is also apparent is that every operator with a significant network should explore the potential of a router off-load strategy. Innovation Observatory has performed analysis that suggests – as far as the cost of expansion capacity is concerned (operators will not want to throw away router capacity they have already bought) – savings in the range of 20–40% are realistic for some operators. Even critics of off-load have produced analysis that shows the use of OTN in combination with routers is cheaper than routers alone.
For operators who are looking to cope with the future growth of transit traffic, router off-load isn’t just the latest flavor in networking – it holds real promise for driving down the cost of operating networks.