Pathway to a Cloud-Ready Wide-Area Network (WAN)

Tighter Planning Cycles for Greater Efficiency with the Evolved Services Platform

In the global geography of telecom, wide-area networks (WAN) are oceans of uncertainty. Resource-constrained and multivendor, WANs produce delays and outages in far-flung and sometimes remote areas, posing a special set of issues that are distinct from those we see in data centers and access networks.

WAN bandwidth is the most expensive bandwidth in the network and failure impacts are large. WANs bear the brunt of traffic growth with a very tricky calculus: underbuild your WAN and jeopardize your brand, but overbuild it and spend your way into oblivion.

Greater Predictability through Ever-Shortening Planning Cycles

To keep pace with these conundrums, you need sophisticated modeling and planning tools, which naturally evolved—in the case of the WAN Automation Engine (WAE)—into an ever-tightening loop of planning, building, and measuring, eventually encompassing SDN.

Longer planning cycles inevitably means over-engineering, over-building and over-hiring. With the Evolved Services Platform’s (ESP) Orchestration Engine, Cisco is shrinking these cycles, and thus reducing the uncertainties that lead to inefficiencies.

Last week I discussed the Orchestration Engine of the ESP in terms of how different components fit in individual domains. Let’s see how to use this framework to plan, engineer, and ultimately automate the WAN to make it cloud-ready.

As the Process Becomes More Automated, a Shrinking Planning Cycle Brings Huge Efficiencies.

The cycle progressively shortens, from years to months, and eventually (with automated, programmable networking) to continuous changes. As this process moves from manual to automated, the network becomes more predictable.

But Why is this Happening Now?

To consider why the evolution of the ESP is now taking this exciting turn, let’s look at some of the underlying technology advances: relevant things are happening at the physical layer, with the network elements, with protocols, modelling tools, and servers.

In short, there is significant progress in every aspect of the network. We now have tools for each of these categories that were either not developed or deployed a few years back:

• Physical Layer: the optical layer is rapidly moving in the direction of supporting alien wavelengths; multivendor capabilities are here thanks to 100G Coherent Optics. The Dense Wave Division Multiplexing (DWDM) Layer is seeing huge advances: colorless, contentionless, omnidirectional, flex spectrum networks where almost all aspects will be controlled by software.
• Network Elements: Vendors are rapidly starting to converge on YANG data models for devices and Network Configuration Protocol (NETCONF) as a standard protocol for interacting with the device.
• Protocols: Work done on Path Computation Elements (PCE) in the early 2000s has been repurposed and redesigned in Path Computation Element Protocol (PCEP), which will allow us to leverage central path computation and program Label-Switched Paths (LSP) remotely. Border Gateway Protocol-Link State (BGP-LS) enables us to collect topology data from the network in a simple way in near-real-time. Segment Routing (which we have implemented in our operating system) enables us to simplify the control plane for Traffic Engineering and make it more accessible to customers with fewer resources.
• Modeling: Modeling tools have become increasingly more sophisticated over the last 10 years and WAE technology has been used in planning some of the largest networks for many years. There is new work being done on multi-layer optimization given the impending ability to control Layer 1 via software.
• Server: Web development tools have been available for many years and promise dramatic improvements in developer productivity for those transitioning functionality from embedded platforms to server platforms. Representational State Transfer (REST) Application Programming Interfaces (APIs) have become universal and there is critical mass in the industry behind open source controllers such as OpenDaylight.

Show Me the Money!

Now let’s turn from time to a business-relevant metric, where the benefits are expressed in terms of the cost savings (and faster revenue generation) that service providers enjoy when they take advantage WAE’s instrumentation and programming in the WAN. We’ll look at the business benefits of a well-planned, well-engineered, and an automated, cloud-ready network.

The Well-Planned Network

WAE allows service providers to respond to overall traffic growth in terms of keeping the whole network stable and not just provisioning to traffic peaks. The savings in doing this can be very significant: in the case of one Internet Service Provider (ISP), they are clearly saving millions in USD per year.

Although under normal operation up to 80 percent utilization of a link is quite functional, the ISP’s capacity planning policy was to upgrade any link that exceeded 50 percent utilization to provide margin for potential failures and consequent traffic re-routing. Using WAE’s what-if scenarios and failure analysis, the customer could now examine the impact of network failures in more detail, and was able to adopt a new policy by only upgrading links that were in danger of dropping packets on single-circuit failures.

The Well-Engineered Network

In a well-engineered network, you should be able to respond to failures or unbalanced traffic in minutes. For instance, using the analytics technologies in the WAE portfolio, an operator gains centralized failure analysis, visualization, and the potential to optimize their network or quickly repair outages. One operator—the before and after picture is shown below—was able to load-balance traffic across transoceanic links, and run links hotter.

Overloaded Circuits are Shown in Purple on the top, Traffic is Balanced on the bottom.

These are the benefits of continuous planning on very expensive links. Based on transoceanic link costs (for example, a single wavelength across 1000 miles of optical infrastructure can lease for $60,000 or more per month), the annual cost savings could be on the order of $20+M USD.  

The Cloud-Ready Network

The above technologies are woven into the predictive analytics capabilities of WAE, a powerful SDN solution that allows you to deploy and optimize innovative new services, such as global load balancing, bandwidth calendaring, bandwidth on demand, and premium network routing. Using WAE, you can dynamically optimize the network as needed, in near-real-time, as a programmable application engine. Operators can respond to traffic shifts of CDN and cloud-based applications in 10’s of seconds.

What might the revenue impact be from this? According to this ACG research report, they can be in the hundreds of millions of dollars over just a few years, when you look at the CAGR of new revenue generating services such as bandwidth scheduling (whether calendared or on-demand) combined with the heightened TCO savings from load balancing and multilayer optimization.

This is the inevitable direction for the cloud-ready network—the 21^st century WAN.

Why do you think WAN is important? Tweet us @CiscoSP360.