1 4 VANILLAPLUS MAGAZINE I OCTOBER / NOVEMBER 2015

I N T E R V I E W

Rob Marson, the vice president of marketing at NakinaSystems, has participated in all three of the VanillaPlus NFVRoundtable Discussions. Here he discusses how the path toNFV maturity has developed during 2015

anillaPlus: How have NFV discussions matured or evolvedover the last 12 months?

Rob Marson: We have witnessed an evolution in the last 12months. This is even evident in many of the discussions we

have had in the VanillaPlus NFV Roundtable series. The discussionsthemselves have mirrored those occurring in the industry in general –there is more discussion about monetisation and operationalisationfactors. We have collectively exited the portion of the hype cycle in whichpeople have understood the basics of the technology, understand someof the benefits. People are now trying to figure out just how can they useNFV to improve their business.

It also feels that we’re past the peak in the hype cycle as there isrecognition that dramatic cost savings resulting from the move tocommercial off-the-shelf hardware may not be easily realized. In fact, insome cases, initial costs will be higher as a result of higher operationslearning curves and complexity. It is only when there is pervasive wide-scale deployment that many of the cost savings benefits will be realized.Ironically, this requires that operational complexities be overcome.

There is widespread acceptance that the benefits of SDN and NFV arenot cost savings, but service agility. SDN and NFV are also now widelyperceived to be enabling technologies, and not unrelated to each other.In fact, NFV is becoming the business case for SDN.

Understanding of the revenue potential from SDN and NFV use casesmust improve. There is an aspect of revenue displacement that will occur,so new value-creation opportunities are needed as communicationsservice providers will continue to face significant OTT competition.

VP: Do you think the challenges relating to orchestration are nowfully understood and being addressed appropriately?

RM: Orchestration is another topic which has evolved over the last 12months. There has been recognisable progress in defining and

Short-term businesssuccess required totrigger larger scaleNFV operationaltransformations

V

�

Rob Marson: We may be reachingthe peak of inflated expectations,and that’s a good thing

1 5VANILLAPLUS MAGAZINE I OCTOBER / NOVEMBER 2015

understanding some of the operational considerations,but there is still a lot more that needs to be done.There is growing recognition service provider networksspan multiple technologies, vendors and geographies,and support millions of end users. In addition, it isunderstood that services will extend across wired andmobile networks and span virtual and physicalinfrastructure. End-to-end orchestration strategiesmust take this into account. A lot of early discussionsseemed to overlook these implications.

The next phase of orchestration will focus on somespecific operational challenges, such as networkconfiguration and change management. Up to 60% ofthe network outages and degradations in a serviceprovider network can be traced back to configurationerrors. Needless to say, ensuring that virtual servers,the VNFs (virtual network functions) themselves, andall supporting systems are correctly configured andstay correctly configured, is crucial to assure servicedelivery and performance. It’s safe to anticipate thatthere will be more industry discussions about defining,maintaining and auditing network and service datamodels.

VP: Lots of proof of concept projects have beenrun in the last year, do you think that trial andexperimentation is nearing an end? Is 2016 theyear in which we'll see virtualisation inproduction networks or might 2017 be morelikely?

RM: Proof-of-concepts have now moved from lab andfield trials to initial – and while some may disagree, Iwould say small scale – commercial implementations.This is another important evolutionary phase that willflush out many more considerations – both technicaland commercial. For many CSPs, short-term businesssuccess will be necessary in order to trigger largerscale and to trigger operational transformations whichare vital. I believe service providers are moderatingtheir short-term expectations about potential extrasavings, recognising that a long journey remains, andthe real business benefits will result in the mid to long-term. This is also likely to be why we’ve seen virtualCPE (customer premise equipment) emerge asprobably the most common and popular initial usecase. It is relatively easier to define, implement andmeasure and in some cases isn’t necessarily a newservice, but rather an improvement and/or more costeffective method to deliver existing services, and growmarket share.

We’ve seen many announcements by serviceproviders of all types and sizes regarding NFV andSDN based service launches. For instance, we’re allfamiliar with AT&T’s stated ambition to virtualise 75%of its network functions by 2020. This means that initialproduction deployments are already taking place.

VP: What do you see as the next priority in thefurther development of NFV?

RM: An area that is lagging somewhat behind iscybersecurity, and we’re finally starting to see itemerge as a key topic for discussion. While data planesecurity services, such as virtual firewall as part of avCPE service chain, is a common use case, securingthe network resources themselves is still largelyneglected. SDN, and its various points of centralisedcontrol, makes the keys to the network kingdomextremely valuable. Complexity introduced by NFV,including multi-tenancy relationships between tenantsand landlords when hosting virtual network functions,an expanded attack surface and the need foradministrative isolation between various domains,highlights that we are only reaching the cybersecuritystarting point.

ETSI’s security working group has started to focus onsome of these topics. With privileged users behindvirtually every recent high profile cyberattack,developing the right identity access managementstrategies for insiders, partners, management andorchestration systems and SDN controllers isnecessary.

VP: The recent SDN & OpenFlow World Congresssaw lots of demos and significant contributions fromCSPs. What were some of the highlights for you?

RM: In my opinion, this year’s event affirms that wemay be reaching the peak of inflated expectations, andthat’s a good thing. For instance, some CSPs openlyexpressed that there is still too much marketing andtoo few deployment-ready products – only roadmapsand long roadmaps at that. CSPs are alsoacknowledging that the initially expected cost savingshave yet to materialise and that there really nostandard blueprints or recipes to follow. There are, andwill continue to be, many different approaches thatindividual service providers will implement in order tomeet their unique needs and circumstances.

We are at the juncture where some of these challengesare starting to be addressed.That’s not to say thatwe’re necessarily sliding headfirst into the trough ofdisillusionment. Few can argue that the benefits ofNFV and SDN are going to be significant and have aprofound impact on the service provider industry. As Imentioned earlier, it was encouraging that some of themore tangible operational considerations are surfacing.There was also a lot more discussion oncontainerisation as another technical approach. Andwhile there is recognition of the effect DevOps(development operations) will have on NetOps(network operations) in the future, SecOps (securityoperations) has been more or less ignored, suggestingthat there may be other, parallel, hype cycles forming.

The next phase of

orchestration will

focus on some

specific operational

challenges, such as

network configuration

and change

management