Accedian is now part of Cisco  |

Avatar photo
By Greg Spear

6 ways operators can succeed with virtualized customer premises equipment (vCPE)

Operators are adopting vCPE to simplify service delivery and maintain high QoE. But, it can be complex to set up. 

Market forces compel service providers to adapt their delivery models, causing them to move away from away from single-function, proprietary hardware appliances and toward virtualized platforms with a strong cloud focus to minimize the costs and limitations of on-premises equipment. There is heavy focus especially on virtualized customer premises (vCPE) deployment as a key, “easy” application for network functions virtualization (NFV).

Easy it’s anything but.

Here are six ways operators can guide their decision-making around vCPE and ultimately succeed at reaching their goals with the technology.

1. Build a vCPE Strategy that Maximizes Value

It turns out that, at least in the short-term, reducing cost is not usually a viable goal with vCPE. (Hopefully, it doesn’t cost more, but the software complexities involved mean it almost certainly won’t cost less, all things considered.) So, don’t make that your justification for deploying vCPE. Instead, think of value, in terms of increasing revenue rather than cutting costs. How will vCPE enable faster time-to-market for new services? Reduce management complexity? Eliminate site visits? Speed up problem resolution? Give customers more control over their service configuration options? Result in more reliable service and better QoE?

vCPE has great potential to improve an operator’s market position, if the focus is on value rather than cost. Differentiation is rapidly becoming all about network flexibility and innovation.

2. Consider the State of the vCPE Market Today

Network virtualization broadly, and vCPE in particular, is new, shiny, attractive, and may seem like a magic bullet to solve many of the market challenges operators are experiencing. But, and this is a bit but, virtualization is still in its infancy (or maybe childhood). Standards and technologies involved are still in development. While many vendors are ready with vCPE products, and early trials are gleaning useful insights, there is still very much a Wild West feel to virtualization.

That means operators who choose to invest in vCPE deployments are early adopters, and take risks in doing so.

It’s a risk worth taking, though, and some of the risk can be mitigated by learning from others’ mistakes.

3. Learn From Others’ vCPE Deployment Mistakes

True, there’s not much of a playbook yet for vCPE. But, from early trials, there is at least an outline of that playbook, and operators can and should learn from each other. Here are a few lessons learned so far.

You can’t eliminate physical customer premises equipment entirely. Some physical service functionality will always remain at the customer premises, even after vCPE is deployed. The form it takes is dependent customer or application-specific requirements.

Service demarcation and network demarcation are not the same thing. These are separate aspects of vCPE. A reliable connection to the customer site is essential. Network layer functions must be virtualized with care to avoid impacting QoE.

vCPE isn’t just ‘software on COTS.’ It comprises of three main elements: a set of virtual network functions (VNFs), a customer premises platform, and a management and orchestration (MANO) software stack. Each has inter-related factors to consider.

Manual VNF provisioning is not a sustainable model long-term. It might be necessary to get started with vCPE, but keep automation in mind for the longer term. NFV orchestrators will mature to meet deployment needs, and operators should plan accordingly.

Avoid injecting new operational complexity. This is challenging—although necessary—
given the many hardware and software options available, and the management changes involved in moving toward a model where service layer and hardware are abstracted from each other.

4. Choose the Right vCPE Deployment Option for Your Network

Every network is different, so there’s no clear-cut answer to the question, “What is the best vCPE deployment model?” These options can, however, be boiled down to three main models.

<click image to enlarge>

Here’s a quick summary of what each option means and its pros and cons.

Localized/uCPE: Appliances at the customer premises are virtualized, replace by NVFI and VNFs to implement network services. The NID function can be a smart module installed in the compute platform, or an integrated VNF. Localized vCPE is attractive for early NFV-based vCPE deployments, because it replaces single-function, legacy appliances with a single NFVI instance. But, this option comes with complexity risks, and is inelastic compared with using centralized data centers to host functions.

Centralized: Functions that were previously hosted on CPE are pulled back into the provider network as VNFs running on NVFI in a data center or edge PoP. This deployment option enables highly effective sharing/pooling of network resources, and simplifies adding compute capacity on an as-needed basis. It also minimizes on-premises hardware to a COTS NID or smart SFP module to demarcate service. But, it requires upfront investment in full NFV infrastructure, and with currently available technology may not be able to meet latency and security requirements for all services.

Distributed/Hybrid: VNFs are placed on an NFVI located either at the customer premises or within the service provider cloud. This is the most flexible option, allowing providers to extend managed services outside their geographic footprint using an overlay network between end locations and cloud. It’s an obvious migration path for operators starting their vCPE migration using the localized model. But, it’s the most complex deployment option and requires sophisticated policy and orchestration for best placement of each workflow’s VNFs.

5. Keep QoE in Mind When Deciding Where Network Functions Should Reside

At some point, decisions about vCPE deployment boil down to answering the question, “Where should VNFs be located”? As outlined above, the answer could be at the customer premises, or centralized in a the data center, or a hybrid of those two. Each provider has to answer this question for themselves, based on use case requirements and (perhaps) customer input. Factors to consider include performance, security and policy, cost, and operations.

A good starting point is identifying which network functions can be effectively virtualized, and which would be better left at the customer premises. How will the VNF placement impact performance?

Operators should take a continuous improvement approach, integrating customer feedback, and both short- and long-term goals into decisions about this business model transformation.

6. Think About vCPE and QoE in the Short-Term and Long-Term

From a purely conceptual standpoint, centralized vCPE is best in the long-term. But, in the short-term, it may not be realistic to jump straight to that model, since technology and standards are still evolving. Many operators will choose a hybrid approach to VNF placement and vCPE deployment, with an eye toward an eventual centralized architecture. To make that possible,  it’s critical that NIDs or their virtualized demarcation and monitoring function equivalents, are put in place now to smooth the transition.

As with any new way of doing things, especially when the change is driven by immediate market forces, it can be hard to pull back and take a big-picture, long-term view. But, with vCPE, this is critical; it’s not just an approach for delivering individual services; instead, it’s a platform for all service delivery. Operators who miss this point risk eventual (and almost certain) peril.

The good news is that, as an early use case for SDN and NFV, vCPE supports the long-term transformation from connectivity provider to value-added services delivery. For many operators, the most straightforward application at the moment is to roll out vCPE-supported services for SMB customers, because they tend to have less sophisticated service requirements than enterprises.

As technology and standards mature, and operational best practices are worked out, rolling out virtualized services to enterprises will become feasible, too.

No matter how you cut it, network virtualization deployment will have a big impact on service provider infrastructure and operations.