Today's VoIP networks are decomposed and disaggregated compared to legacy voice networks. What was once a box is now several software functions and what was once single vendor is now a multitude.
While network functions virtualization (NFV) is central to success in network transformation, when it comes to VoIP, building the network cloud is difficult for many service providers for a number of reasons, including cost, complexity and time to market. Could NFVaaS be the answer? (See NFV vs. the Cloud.)
Network disparateness results in challenges in both the construction as well as operational support and troubleshooting. The design and integration of numerous systems and management of a multi-vendor environment demands expensive expertise and significant time. This was true in the first steps of VoIP and got overly complicated with IMS.
NFV complicates configuration and management further because the software is divorced from hardware and there is a new virtualization, orchestration and management layer inserted in between. To bring a single functional element into the network, it involves several vendors (server, virtualization layer, telco software/VNF, MANO, OSS/BSS). That's lots more stuff that needs configuring. Ultimately it means more vendors and raises questions on who owns a problem when it arises (i.e., many throats to choke).
By some counts a basic IMS call could comprise 136 messages across 17 functional elements and involve six different protocols. NFV adds even more functions and more interactions which may be inherently better in the long run, but there are more moving parts, more chances that components can get out of synch and upgrades and troubleshooting may be a nightmare. And adding to this complexity is that some of these inter-function and management interfaces are not matured or defined.
This complication throws a wrench in the plans for lower opex and service agility. (See Telecom Faces Virtualization Pricing Issue.)
The complex operational and support environment is one of the major drivers for cloud sourcing voice as opposed to rebuilding a voice network (NFV or otherwise). Done right, cloud sourced NFV voice consolidates all the voice functional elements into a single elastic, fully managed and horizontally integrated solution. It's more like NFVaaS for VoIP, or as Tom Nolle recently wrote about: VNFaaS.
This allows the solution platform to be managed as a single system with flow-through provisioning and automation versus managing a collection of vendors and virtualized functions. And because this is done in the cloud there is no risky and expensive platform upgrade required. If it's cloud-based, the focus is on product and customer management, not technology tinkering.
Horizontal integration of the functional elements means a single management interface is provided to define, manage, provision and troubleshoot voice services. This single view reduces operational costs, facilitates scalability and accelerates problem resolution. It's the horizontal integration (not just coordination and orchestration between VNFs) that's key to simplifying operations.
At the end of the day the backend technology only matters to the extent that it provides voice service at lower cost, higher quality and facilitates easier management and provisioning. Configuring a subscriber should be as simple as setting up a Facebook or Gmail account instead of requiring special domain knowledge and access to numerous systems and interfaces.
NFV is essential for providers this decade, but I don't believe building the network cloud is the right move for many service providers when it comes to VoIP. Instead, cloud sourcing NFV, or leveraging the already built cloud as the next-gen voice solution, is a smart way forward because the cloud is a way that NFV technology can be consumed and used.
While NFV promises long-term cost savings and service agility, it's not a magic wand that brings about fully compatible, horizontally integrated, zero-effort components. The benefits of NFV will only be realized after a lot of effort and expense. It's tough to "fail fast" when the architecture takes five years to solidify. So what makes sense to you given that transformation is needed now?
— Kevin Mitchell, Vice President, Marketing and Product, Alianza, special to The New IP