Friday, February 27, 2015

Blueprint: vCPE Evolution and NFV-Powered Network Assurance

by Scott Sumner

With traditional business services shifting toward federated cloud connectivity, communication service providers (CSPs) have set a new course for delivering these critical services. Virtualizing customer premise equipment is a key part of their strategies—vCPE is a cost effective, proven technology approach that’s already showing significant returns.

The Promise of vCPE

Business services are not what they used to be. Enterprises increasingly are migrating to public and private clouds, and connectivity to these hosted resources is ever more critical for business success.

Margin and revenue threats to communications service providers (CSPs) lurk around every corner: public cloud platforms, over-the-top (OTT) wide area networking (WAN) technologies, and cheap, dumb pipes, to name just a few.

CSPs can win big, though, if they are able to cost-effectively deliver performance assured multi-cloud, multi-service connectivity. In this context, a vCPE strategy is becoming a critical component in a competitive service delivery model. For example, Colt Technology Services achieved remarkable benefits from its broad NFV/vCPE strategy intended to all but eliminate on-premises appliances; over a three-year period, virtualizing routers resulted in more than 70% CapEx savings.

Why NFV-Powered vCPE?

Data center connectivity services face intense cost pressure, yet cannot be delivered in a way that sacrifices quality of service (QoS) since this connectivity is the lifeline between enterprises and their business-critical infrastructure.

When a CSP adopts a vCPE strategy, their main goal is to use NFV to replace as much equipment as possible at the customer premises with virtualized equivalents, thereby reducing CapEx and increasing agility.

Many vCPE deployments already use intelligent NIDs to perform critical customer premises edge functions like Layer 2 and 3 QoS mapping, hierarchical traffic conditioning, service OAM, and performance monitoring, allowing routing and other L3+ functions to be located and/or virtualized deeper into the network.

But, while this approach is still a highly cost-effective method to deploy services in a multi-tenant environment, unit cost can lengthen return on investment (ROI) when serving a single customer

Curiously, CSPs often refer to the last remaining piece of equipment—a demarcation unit or intelligent edge appliance—as “the vCPE”. In this context, CSPs are taking virtualization to its logical conclusion, by applying NFV to the vCPE appliance itself .

By using NFV-based vCPE appliances, CSPs can not only realize significant CapEx and OpEx savings, but also increase their performance visibility, greatly decrease deployment time with customer self-install capabilities, outmaneuver competitors with higher cost structures, and compete on performance rather than price.

But, in the real world, what does it takes to replace racks of traditional CPE with centrally managed software?

A Closer Look: NFV-Powered vCPE Approaches

Adopting an NFV-powered vCPE strategy can only be successful if the technology involved seamlessly integrates legacy and virtualized approaches into a unified delivery platform. When that’s true, CSPs gain a smooth migration from existing infrastructure, within existing operational practices.

One approach to using vCPE technology is virtualizing all computationally intensive performance assurance functions—such as report generation, test session sequencing, and statistics—and retaining the minimum possible hardware to achieve the equivalent precision and line-rate processing offered by standalone edge appliances. Programmable smart SFPs or compact GbE modules can provide the local processing required—with the full feature set of NIDs realized when complemented by performance assurance virtual network functions (VNFs) hosted in a central controller.

Another approach (referred to here as vmCPE for “virtualized module CPE”) is to fully virtualize all vCPE functions to a software-based appliance running on COTS hardware. This vmCPE appliance can use the same catalog of centrally hosted VNFs, if its architecture permits.

As with all technology decisions, there are trade-offs between these two NFV-based vCPE approaches. The first approach (compact hardware modules) is more precise but not always practical; some locations—such as services terminating on a virtual machine—cannot be monitored using any physical device.

The vmCPE approach overcomes those visibility challenges and can also collect related VM metadata (CPU load, memory usage, etc.) to help isolate network issues arising from NFV infrastructure performance variation. But, vmCPE appliances cannot realize the same order of precision or packet-processing efficiency as hardware-based modules, because CPU load, varying degrees of access to underlying hardware, and the asynchronous, free-running clocks in virtual machines prevent deterministic time stamping and limit accuracy.

Regardless of approach, the ability to host and move VNFs to diverse locations needs to be supported by an effective NFV-powered vCPE solution deployment strategy. Various NFV-based implementation options emerge when standalone vCPE infrastructure is broken down into its logical components—allowing CSPs to employ variants that address site-specific requirements while retaining unified control and visibility over all service endpoints.

What’s Ahead for vCPE?

Clearly, the vCPE is becoming a vital component in the connectivity service delivery path and its lifecycle. This makes migration to NFV inevitable. The best, least painful way forward is to unify a variety of programmable vCPE options, giving CSPs the benefits of extensibility and cost savings via orchestrated virtualization implemented in an open, real-time architecture. Such an approach needs to be carefully designed to ensure solutions are highly extensible, scalable, and programmable, and openly interoperate with other platforms and functions.

vCPE implementations will increasingly require a mix of legacy, NFV-powered and fully-virtualized approaches, and will use software defined networking (SDN) standards to build a centrally controlled, highly-scalable distributed networking fabric that harmonizes physical and virtual instrumentation. Result: CSPs won’t have to manage standalone, NFV-based and fully virtualized vCPE infrastructure separately; these will seamlessly interact.

About the Author

Scott Sumner is VP of solutions marketing at Accedian Networks. He has extensive experience in wireless, Carrier Ethernet and service assurance, with over 15 years of experience including roles as GM of Performant Networks, Director of Program Management & Engineering at MPB Communications, VP of Marketing at Minacom (Tektronix), and Aethera Networks (Positron / Marconi), Partnership and M&A Program Manager at EXFO, as well as project and engineering management roles at PerkinElmer Optoelectronics (EG&G).   He has Masters and Bachelor degrees in Engineering (M.Eng, B.Eng) from McGill University in Montreal, Canada, and completed professional business management training at the John Molson School of Business, the Alliance Institute, and the Project Management Institute.

About Accedian Networks 

Accedian Networks is the Performance Assurance Solution Specialist for mobile networks and enterprise ­ to­ data center connectivity. Open, multi­vendor interoperable and programmable solutions go beyond standard ­based performance assurance to deliver Network State+™, the most complete view of network health. Automated service activation testing and real­ time performance monitoring feature unrivalled precision and granularity, while H­QoS traffic conditioning optimizes service performance. Since 2005, Accedian has delivered platforms assuring hundreds of thousands of cell sites globally.

Got an idea for a Blueprint column?  We welcome your ideas on next gen network architecture.
See our guidelines.