Q&A: NFV Adoption for Mobile Operators - Benefits and Challenges

Posted by Mae Kowalke on Thursday, August 13, 2015 with No comments

Mobile network operators are increasingly adopting network functions virtualization (NFV) as they move toward virtualization and automation in their networks. Here, Accedian's VP of Marketing Scott Sumner answers some questions about how this adoption is going, the challenges involved, and what operators hope to achieve. 

Why is NFV adoption increasing?

Network operators are primarily interested in delivering services and new, revenue-driving applications more quickly, so that they can compete on a level playing field with the software-centric offerings of major cloud providers like Amazon and Google. A closely-related need is the ability to move to open networks running on COTS servers, to reduce cost and avoid proprietary vendor lock-in.

What's the first thing a mobile operator should do before moving to NFV?

Network complexity will increase dramatically in the migration to NFV-based infrastructure. The network will become more dynamic. Performance visibility across both physical and virtual domains is necessary to (1) make sure virtualized infrastructure (NFVI) supports service chaining performance requirements, (2) the virtualized network functions perform as well or better than the legacy one, (3) rapidly troubleshoot any issues affecting user quality of experience, and (4) allow SDN controllers to dynamically optimize performance across the network using a real-time view of QoS as an input.

What role does software-defined networking (SDN) play in the moved toward virtualized, software-based solutions?

As NFV is deployed, it will be common to spin-up virtual networks and services that will coexist in the same network, in different network ‘slices’. Each will have their own requirements from the network to support the various applications each will carry. SDN controllers have a critical role in ensuring all network resources, physical and virtual, are being used as efficiently as possible, while maintaining optimal performance as required by each service or virtual network it controls. This level of complexity cannot be managed by existing routing techniques, as it requires a central view of network-wide performance (the network state) to make routing and prioritization decisions.

What's the best way for a communications service provider to approach the migration to NFV: one function at a time, all at once, or something in-between?

NFV can be implemented in phases, starting with areas where agility and cost savings are easiest to quantify and provide the fastest return. Most service providers see virtualizing the customer premise equipment (vCPE) as a clear target for NFV, and mobile operators are applying the same logic to their backhaul networks, where the small cells and macro cells would replace the equivalent of customers in enterprise connectivity scenarios.

What are the biggest challenges related to virtualization network functions?

One main difference between physical network elements and their virtualized counterparts comes in performance. ASICs and dedicated hardware offer unique line-rate packet processing capabilities that cannot be fully replicated in COTS servers, including precise time stamping (for critical measurements such as latency), and transmission scheduling (for accurate diagnostics and test stream generation), in addition to the inability to address full line-rate traffic in many cases.

For certain critical applications, such as performance monitoring and network sync, small amounts of dedicated hardware will still be needed to ensure QoS and QoE are not sacrificed. Using NFV to minimize this hardware footprint is an elegant technique that preserves hardware performance at the cost and agility targets NFV aims for. Smart SFPs, that can be programmed to perform many functions, are an example of a lightweight approach to maintain a balance between hardware and software solutions.

What is the biggest hidden benefit of NFV?

The ability to do things that a service provider was never able to do before--because of cost, deployment effort, or operational practices--is a place where NFV may deliver significant, but unexpected, benefits.

Example: virtualized instrumentation. Before NFV-based solutions were available, it was impractical to gain full network monitoring visibility across and entire network in a cost-effective, uniform way. Now a full network (100,000s of cell sites, for example) can be instrumented within a few weeks, unifying the view of the entire multivendor network infrastructure. A nationwide deployment Accedian recently worked on with SK Telecom in Korea is an excellent example of this type of benefit.

What role does the cloud play in NFV migration for network operators? 

In recent years, the enterprise-to-data center connectivity landscape has dramatically transformed, from mostly private site-to-site wide area networks (WANs) to data and applications hosted on distributed public and private clouds. More than half of enterprise workloads are now crunched in the cloud. This will have profound effects on mobile traffic, as well, as increasingly enterprise apps are accessed from mobile devices.

Mobile network operators are in a unique position to offer these dedicated, reliable, flexible connections with sufficient availability, low-latency, configurability—the cornerstone of hosted services and applications.

Providers can offer value-added access services by using their differentiated network assets (including Internetwork Packet Exchange (IPX) private, confederated networks) epositioning themselves as cloud connection providers, delivering an assured pipe that combines dedicated connections to different cloud colocation points, provider-furnished managed services (such as telephony and security), and the internet.

As this market continues to emerge -- coupled with the emerging opportunities in wireless enterprise access connections -- the cloud will become a significant opportunity for mopbile network operators if they are correctly positioned.

How are network operators rationalizing their move to NFV?

Many must justify their deployments on cost savings, but in parallel are expecting to show immediate gains in revenue growth as new products are offered that leverage this new, more agile infrastructure.

What should operators expect from the changes associated with migrating to NFV, and will reality meet expectations?

The results operators achieve from this migration will largely be determined by their approach. Those that plan ahead carefully and go into the transition with eyes wide open are already showing significant benefits as they move towards a goal of fully automated network infrastructure. Those that apply NFV in a particular application, but fail to pull it into a holistic plan, will still see benefits, but not to the degree of those that unify NFV, SDN, and virtualization into a new way to build and manage a network. Time will tell.

Will 2015 become known as the Year of NFV?

For many operators, it already is. It certainly is a year of wide-scale deployments in vCPE, virtualized mobile infrastructure, and many applications--including emerging ones such as Network Platform as a Service (NPaaS). As new revenue streams open up, we’ll return to this year where the foundations were being built, and see this year as a turning point in networking and services
Categories: , ,