By: Joaquin De La Vega Gonzalez-Sicilia, Alcatel-Lucent Sr. Manager, Cloud Consulting Services
I still remember with great excitement how, in October 2012, a group of network operators published a whitepaper that coined the term Network Functions Virtualization. This announcement validated a vision that we had been promoting under the name of “Virtual Telco” for more than two years. The telecommunications world had decided to start a fascinating journey towards the cloud, and we were already in the game with a product. What I could not imagine is how fast things would move.
Since then, many players have announced their plans, and many of them have shown some kind of functionality addressing different aspects of NFV. CTO teams have been leading the discussions around performance and functionality, but until now, nobody had tackled one of the major questions: Will the NFV business case fly?
We had begun internal discussions around the business case to support the NFV value proposition and had finished a framework when an opportunity arose to work on a virtual DNS business case with a customer. In the course of a DNS demo presentation on top of the Alcatel-Lucent CloudBand™ NFV Platform, one of our tier-one service provider customers pointed out that their large fixed network DNS deployment was at end-of-life and they were planning to replace it. The discussion quickly focused around the question of whether migrating to NFV with CloudBand would be more beneficial than replacing the old servers with new ones, but keeping the traditional mode of operation. To help with this decision, we offered to develop a joint business case comparing both scenarios. The customer agreed, and within two weeks I was working with our client’s DNS lead at their premises.
A first challenge was to develop a suitable methodology for the analysis. We decided we had to map out the main processes in DNS operations in a way similar to Gantt charts. That was a laborious task, but was key to capturing the differences induced by NFV. We analyzed how the processes were performed today, and how they would change to fit the NFV model. To quantify capacity requirements and get credible numbers, we tested DNS loads on a real CloudBand node in the customer Lab. As a result, we had to change our initial estimations, which turned out to be too conservative. Finally, after about ten weeks of joint effort, we achieved what looked like a sound business case. It had been a great challenge but the effort had paid off.
The work delivered two results we had not necessarily expected. Even a simple application like DNS can benefit clearly from running on an NFV platform. Processes such as scaling, software upgrading and healing are greatly simplified, which increases agility and significantly lowers total cost of ownership. Secondly, the results were clearly positive with a single application running on the NFV infrastructure, that is, service providers can start small on the road to NFV. It is not necessary to deploy many virtual network functions all at once. Of course, sharing the infrastructure will provide the full benefits.
Looking at the demand for presentations both from partners and customers, I can now see how important this exercise was to better understand the economic value of NFV, and how it contributes to the bottom line. Great days are ahead of us. I cannot wait to see the first wave of NFV deployments that will change the telecoms world as we know it today.
For additional information, read the ‘NFV Insights Series: Business case for moving DNS to the cloud’.
