Christos Kolias is a senior research scientist at Orange Silicon Valley (a subsidiary of Orange), leading its software-defined networking (SDN) project. Christos is a founding member of the NFV group within ETSI and he is also chairing the ONF’s Wireless & Mobile discussion group. He has lectured on OpenFlow, SDN and NFV at several events. He has more than 15 years of experience in networking and holds a Ph.D. in Computer Science from UCLA.
SDxCentral: What excites you about NFV? What is your view on the potential impact of NFV?
Kolias: “NFV presents a fresh look at the operator’s network, which was very much needed, and this is happening in the wake of software-defined networking. What I particularly find fascinating about NFV is its vision to change the landscape of the telecom/service-provider industry (and beyond — for instance, in the data centers) and the coalescence, in a way, of an ecosystem, in a very short time I must say, around this initiative, totally embracing the concept, its prospects, and the promises it can deliver.
I think the impact will be measured in various areas and at different scales. It will bring flexibility, agility and automation and a much faster time-to-market cycle, where the latter is something that we, as operators, lack today. I am also hopeful that it will be able to introduce new services and functions that we could not think of before, and not simply move into software current network functions. So it promises to change the landscape in many ways and forms with far-reaching consequences.”
SDxCentral: What’s Orange’s view of how NFV will benefit you and your customers?
Kolias: “Same as for other operators; we are very excited about the overwhelming opportunities and tremendous benefits that NFV presents, the opportunities to innovate and strategize about new services and streamline TCO [total cost of ownership]. The specific benefits remain to be seen once we start rolling out NFV-based services.”
SDxCentral: The NFV initiative is not a dedicated organization like the ONF. Can you describe what it is and how it is funded?
Kolias: “NFV is an industry specifications group (ISG) that is under the European Telecommunications Standards Institute (ETSI), who are very gracious with providing us with lots of secretarial support and guidance . If you are a member of ETSI, you can automatically become a member of our group; if not, all you need is sign a participant’s agreement and pay a nominal fee for attending our face-to-face meetings.”
SDxCentral: You have a role within ETSI on the NFV committee — can you describe what it is?
Kolias: “I’ve been fortunate enough to be a founding member of this initiative and I am handling the group’s inbound and outbound communications and public relations. I have been also nominated as the liaison between ETSI NFV and ONF, tasked to identify any collaboration and synergy opportunities between the two groups — and they think there are plenty such opportunities. The SDN initiative presented somewhat an impetus for the formation of our group that essentially dates back to the confines of the 2012 Open Networking Summit (ONS) in Santa Clara, Calif.”
SDxCentral: Is NFV truly complementary to SDN, or are they completely orthogonal?
Kolias: “We do view SDN as highly complementary and synergetic to NFV. I have not seen, myself, cases where the two are completely orthogonal, other than: In order to implement the virtualization of network functions, one does not necessarily need an SDN approach. While SDN is characterized by the clean separation of the control and data plane and the programmatic nature of it, NFV is about moving network functions into software, so clearly the common denominator is software. I see management and orchestration as the most prevalent common key theme in both NFV and SDN technologies.”
SDxCentral: When do you think we’ll see NFV standards coming out?
Kolias: “We have an internal mandate to wrap up our activities by beginning of 2015, as ETSI ISGs have, by definition, a short lifespan. The objective of the group is to provide a requirements framework, requirements recommended by us, the operators. That set of requirements could lead to the emergence of standards at a later stage — but this is not a currently a goal of this group.”
SDxCentral: What’s your view of vendors that claim NFV compliance already today?
Kolias: “Although we do not necessarily claim that NFV life started with ETSI, some of the operators and vendors had been already working on such concepts prior to the group’s formal establishment/formation. If one is able to virtualize network functions, essentially implement them on software running on commoditized servers, I would think they are within the scope of NFV. We certainly want to avoid the concept of “NFV-washing!” (That will subsequently require some “NFV-drying.”) We will need to carefully look into what “NFV-compliant” means, especially in the absence of any standards or certification process today.”
SDxCentral: What are the first functions you would expect to see virtualized?
Kolias: “We are currently working on delineating and elaborating on a number of use cases that NFV would be not only applicable and feasible but also implementable. As such we have identified vCDN, vCPE, vSTB, vRGB, vEPC, vEnterprise — note that most are preceded with a ‘v’ to denote the virtualization aspect. Other fields such as DPI, load balancers and security functions (firewalling, IDS/IDP) are also on our radar screen. We need to clarify here that our focus, as a group, is not on virtualizing individual boxes, but rather to provide the necessary constructs, primitives and overall architecture that can lead to the virtualization of the network functions.”
SDxCentral: Are you already running initiatives internal to Orange around NFV? What’s been the success so far?
Kolias: “In Orange, being one of the group’s seven founding members, we are very keen on working on different aspects of NFV. Given this is a common effort, areas such as CDN, home networking, IMS and SBC, and the mobile packet core present certain interest and promise. More specifically within Orange Silicon Valley, we are looking into the vEPC use case, setting up a proof-of-concept (PoC) testbed for experimenting with different solutions (I personally believe that mobile network functions virtualization — mNFV — appears very attractive). NFV can play an important role in designing the next generation PoP.”
SDxCentral: What do you see as barriers to quick adoption of NFV, and how fast do you see NFV actually being deployed in carrier networks? One to two years, three to five years, or longer?
Kolias: “Certainly the concept of NFV is very appealing and optimistic. Integration into the existing infrastructure and operations could be disruptive and hence challenging, in some cases. We are already, within ETSI NFV, working on PoCs. The interplay and/or integration with the OSS/BSS will be critical.
Looking forward, the emergence of an end-to-end architecture could significantly simplify operations. Although this effort is about virtualization (hence about software), it would be interesting also to see how silicon vendors such as Intel, ARM and also Broadcom, Marvell, Tilera, Cavium, Freescale, Netronome and others will respond to the NFV efforts — we are certainly open to new ideas and we hope the best emerge.
Taking into account current investment in equipment and being mindful of the depreciation cycles associated with hardware, the adoption of NFV could vary, not only from operator/service provider to operator/service provider but also within an operator’s network as priorities could be given to certain areas that would depend on how mature (or how hemorrhaging, in terms of costs, revenues, and performance) that area is. NFV looks certainly ripe for new, green-field types of deployments or in cases where you want to dynamically add resources (as NFV can be supplemental to existing infrastructure). Hence adoption cycles could also vary, but I would say that we, as a community, are ready.”
SDxCentral: Thank you for your time, Christos!