Is successful orchestration the key to successful NFV implementation? The short answer to that is Yes … and No, as Jeremy Cowan found when talking to experts in network functions virtualisation (NFV). As you might expect the answer is a little more complex.
Before we get to orchestration, the senior network architect at carrier Colt Technology Services, Javier Benitez, believes that we need to address another matter. Prior to orchestration, he says, the VNF (virtual network functions) need to deliver the required network functionality at the required performance, otherwise NFV will be a ‘no go’. “If by virtualising existing services CSPs aren’t able to offer the same level of performance and meet the SLAs (service level agreements), an NFV model won’t work. It’s critical to make sure that NFV will deliver the same functionality and performance that CSPs are providing today.
“However, having said that, NFV will never succeed, even if the previous requirements are met, if orchestration is not part of the solution. In the same way that cloud computing can’t be productised without full orchestration around it, NFV can’t go into real production environments if the orchestration piece is not fully in place,” Benitez insists.
Ken Dilbeck of TM Forum is unequivocal. “Yes. Orchestration will be key, not only at the physical configuration level, but also at the service enablement level. The only way to realise the full promise of NFV is to put in place a coordinated orchestration capability across the end-to-end environment. Throw in the orchestration of a multi-tenant environment spanning multiple administrations, and it is pretty obvious that a very sophisticated orchestration capability (or something like it) will be vital.”
Not the only factor
Tony Poulos, Enterprise Business Assurance market strategist at Portugal-based WeDo Technologies, cautions that, “While successful orchestration is definitely a key factor, it is not the only critical area that needs to be addressed. Successful implementation will go beyond reducing dependency on dedicated hardware-based appliances.
“The virtualisation of network functions will greatly reduce the complexity associated with introducing new services across the entire network, enabling the physical network and virtual network to interconnect functions such as optimisation, firewall, and deep packet inspection (DPI) to achieve service chaining. This is where orchestration becomes critical. It should also be noted that the variable nature of costs and resources in a CSP environment makes orchestration a significantly complex task in their networks in comparison to traditional data centre environments.”
Peer 1 Hosting’s senior architect Gary McKenzie, tells VanillaPlus, “It obviously depends on what you use (orchestration) for but for organisations with ambitions to scale their solutions, good orchestration is going to make or break their NFV deployments.”
“Of course, the requirement for good orchestration isn’t unique to NFV and isn’t new either, large scale communication service providers (CSPs) have long known that ‘zero touch’ provisioning and replacement are key to a reliable and positive customer experience.” Mackenzie believes that orchestration is really just the next step on that journey.
It’s also about money. According to Béatrice PiquerDurand, a vice president at Ipanema Technologies, “From conversations with our partners, the key point for deeming an NFV implementation successful ultimately boils down to reducing transformation costs. In the telecoms industry, change is an expensive process, and anything that can be done to reduce these costs is of economic benefit. Any business model that is able to rapidly adapt to changing demands is going to use existing infrastructure more efficiently.”
The money will likely stem from automation, simplification, and maximisation of resources, she believes.
Another risk with NFV
Shaul Rozen, Amdocs’ director of product strategy, is adamant that successful orchestration is “most certainly” the key to successful NFV implementation. “When introducing new virtual use cases one at a time, service providers risk retaining the current state of multiple silo networks that require numerous orchestration and management systems. This lessens the opportunity to reduce complexity and take advantage of full cloud elasticity and agility benefits. The transformation needs to be done in a thoughtful way that mitigates this risk through advanced planning and intelligent, automated orchestration.”
NFV benefits lie in the ability to automatically fulfil and manage network services lifecycle across network
segments and technologies, says Rozen, such as access and aggregation networks, core-network data
centres and on-premises demarcation points. Scaling the NFV promise requires an operations solution that enables service providers to rapidly define new service models and fulfil them instantly upon demand. Key requirements of such a solution include:
• Dynamic service design and automatic fulfilment
• Automatic management of tens of thousands of instances in real time, over a distributed and multi-vendor infrastructure
• Support for services spanning virtual and physical resources
• Integration with business support systems (BSS) and operational support systems (OSS), and
• Integrated policies and analytics for optimal service performance.
Orchestration of the overall network will be key to the success of an NFV implementation. Ben Parker, principal technologist atGuavus has no doubt about that. “Virtual networks allow the operator to change operational models. They can enter new businesses based on new models of operation. “They will also be able to offer a different quality of experience (QoE) on different networks to different people. With the competitive environment for telecoms becoming more and more complex, each operator must be able to adapt quickly both to offer new products, and discontinue poorly performing ones more quickly,” Parker concludes.
Yes, successful orchestration is the key to successful NFV implementation, according to Sameh Yamany, CTO and VP of Mobile Assurance & Analytics at JDSU, “especially when it comes to the Phase Two of self-configured NFV networks. When the network becomes self-configuring, it has to make ‘good’ orchestration decisions – good in the sense of delivering improved business results such as profitability and/or improved customer experience.”
Yamany believes the quality of these self-configuring orchestration decisions will ultimately determine the success of the NFV implementation, and this in turn will be down to the quality of the performance data and policy control functions being used to formulate these decisions.
The biggest challenge
The conclusion is clear to Gerry Donohoe, director of Solutions Engineering, Openet. “The full potential of NFV can only be achieved with the correct orchestration capability. The ability to achieve faster time-to-market, targeted service introductions, and elastic scaling can only be delivered if all of the functions can be automated, such as VM Instance management, and integrated monitoring and metering of workloads. This still remains the biggest challenge, and greatest enabler, for delivering on the potential of NFV.”
I’ll leave the last words to Rob Marson, Nakina Systems’ vice president of Marketing. “While many benefits associated with cost improvements from leveraging commercial-off-the-shelf (COTS) servers are achievable without orchestration, value creation resulting from new service innovation and velocity will not be possible without successful orchestration strategies.”
“In addition, a service-oriented view to management and orchestration is necessary. This implies that orchestration and correlation of network data integrity be extended across all networks, and that physical and virtual environments be bridged. A common mediation and abstraction layer is required in order to bridge various environments together and to enable management, orchestration, analytics and assurance systems to communicate to networks in a consistent and holistic fashion. Otherwise,” says Marson, “the full benefits of NFV resulting from increased automation will not be realised.”
Read more on VanillaPlus.