Eurocomms.com: What stage do you believe SDN is at now – hype or reality – and how will it evolve over the next 12 months?
David Noguer Bau:SDN is still at the peak of the hype curve, but there is a new technology that is scaling positions rapidly and it is Network Function Virtualisation or NFV. Obviously, the two technologies are interlinked so looking at the use cases, NFV will help operators to adopt SDN quickly.
There are two main areas of application for SDN, simplifying the virtualised datacenter and delivering advanced services in the WAN. The industry is closer to making the virtualised SDN datacenter a reality, while the WAN service delivery piece is still hype, but here NFV will help.
Each vendor has their own view of SDN – how committed are you to ensuring a more coherent industry vision?
Until last year the industry was looking at SDN as an equivalent of OpenFlow. Juniper has contributed to expand this view and get SDN back to its original meaning: Software Defined Networking. It is much more than a single protocol. With this new vision, new use-cases have emerged, beyond the virtualised data center. Juniper is fully committed to SDN. Earlier in the year we shared our 6-4-1 vision, and during the second half of 2013 we launched the SDN controllers: Juniper Contrail and OpenContrail.
Are operators moving quickly enough to implement SDN?
Operators are starting to move. We see some ignoring it [and some] only interested in learning about it at this stage. Others have started limited tests in their labs. Then we have those [who are] looking at specific use-cases that can be implemented over their existing infrastructure using SDN. I have recently seen cases where engineers [at operators] ask themselves: “How would I solve this problem using SDN?” The implemented solution might not be SDN, but it influences the final result.
There is nothing like being able to solve real problems to get [operators] moving faster. Our focus right now is helping our customers to work on real use-cases.
What is the biggest challenge that operators need to focus on to ensure SDN is a success?
In a world where new operators could be created from scratch, SDN would have a much bigger role. A big challenge is the infrastructure currently deployed; it doesn’t easily integrate with SDN. But a bigger challenge appears when operators keep on building the new networks in the traditional way – it is a missed opportunity. It is also true that at this stage not many operators want to go for a proprietary implementation, so open solutions will be preferred. But we are finding C-level executives [who are] interested to understand how should they change their organisations to be able to embrace SDN much faster.
Some operators have cast doubt on whether SDN can really engender revenue opportunities alongside anticipated cost savings. What would you say to convince them that this is possible?
The current situation for operators, without SDN, isn’t optimistic - year-or-year aggregated revenues are in decline. The SDN and NFV use cases that Juniper presents are designed to facilitate the creation of innovative services over a simplified infrastructure. This would minimise the cost of providing such services and at the same time it opens new revenue opportunities. Without SDN and NFV those services could also be launched but at a higher cost, so [the operator] won’t be competitive. A couple of examples to illustrate those innovative services:
· Telco Cloud services: Operators find it hard to compete and differentiate their cloud offering in the market. SDN not only helps them to efficiently operate virtualised data centers, but it opens the door for them to offer end-to-end SLAs.
· Value added services to existing VPNs: VPNs still provide a strong revenue line to operators although the competition is tough. To enhance this service, operators had to invest in new CPEs for their customers. With NFV, operators can now provide additional features easily by inserting service chains in their traffic flows. With the help of SDN, this process can be fully automated and self-provisioned.