As the interest in SDN continues to increase, we are also seeing growing interest in a related network architecture: Network Functions Virtualization (NFV). Organizations that are evaluating the viability of NFV have to make a key decision early on: Should they go forward with a piecemeal approach or should they go for a big bang approach?
NFV is getting a lot of attention for the same reasons that SDN has gathered so much attention – both hold the promise of reducing cost and making IT more agile. While the tech is still in its infancy, the early indications are that some players are taking a tactical approach (a.k.a., a piecemeal approach) while others are taking an architectural approach (a.k.a., a big bang approach).
Is a Piecemeal or Big Bang approach best?
Telecom Italia is an example of a company that is taking a piecemeal approach to NFV. As part of this approach the company participated in a proof of concept (POC) called Closing the loop: Data-driven network performance optimization for NFV and Self Organizing Networks (SON). This POC demonstrated how to build a closed loop using key performance indicators such as network performance, customer experience, and service quality to enable dynamic network changes, optimization, and self-healing. As part of the POC, configuration and performance data was collected in a mobile network and then analyzed to understand where problems existed or where there was potential for improvement.
Companies that take a piecemeal approach typically focus on one, or at most a small number of use cases for which they see clear business value, and they typically participate in POCs of the type mentioned to demonstrate the viability of the possible solutions. They also work with vendors to ensure that the necessary functionality will be available, and they work with the appropriate groups inside of the company to ensure that the functionality can be implemented and managed and that it doesn’t present any undue security risks.
In contrast to focusing on rolling out solutions for just a few use cases, when a company takes a big bang approach to NFV they decide on an architecture, and possibly on some key enabling technologies and products, that it will use to support any and all NFV use cases. AT&T is an example of a company that is taking a big bang approach to NFV, as evidenced by its Domain 2.0 initiative that it announced about 18 months ago. One of the goals of that initiative is to migrate AT&T businesses to a multi-service, multi-tenant platform that replaces or augments its existing network elements. In a blog post published in December 2014, John Donovan, Senior Executive VP of AT&T stated that “After a solid start earlier this year, we’re planning to kick our transition to a software-centric network into high gear in 2015. In fact, I’m putting a line in the sand today: our goal is to virtualize and control over 75 percent of our network using this new architecture by 2020.”
Short-term Rewards vs Long-term Risks
There is nothing inherently right or inherently wrong about the piecemeal or the big bang approach. Companies that take a piecemeal approach will often get short-term rewards from the use cases that they move into production. They do, however, face a long term risk of having multiple instances of NFV-related services that have little, if any, commonality and hence may not achieve the promised cost savings and agility. Companies that take a big bang approach are facing big risks up front, in terms of moving forward with a specified approach and a set of technologies in a time when architectural approaches and technologies are changing rapidly. If these companies do a good job of implementing a common platform for all NFV-related services, however, they will likely reap the potential cost and agility benefits.
The ultimate choice of piecemeal or big bang may well come down to the culture of the organization. Does the company function better in a bottoms up mode, so the piecemeal approach would work well, or is it a top down type of company and so the big bang approach would work well?