It seems like everywhere you turn, telco professionals are talking about network function virtualization (NFV) and software-defined networking (SDN) — and deservingly so. NFV and SDN promise transformations that will improve network efficiency and customer satisfaction, just to name a few benefits. As such, mobile operators have begun to take notice and are directing efforts toward incorporating both technologies into their network plans.
With this in mind, Cyril Doussau and I conducted a webinar that explained how mobile operators can ensure a smooth transition to NFV and SDN and reap these benefits. Webinar participants asked a lot of intriguing questions, some of which we weren't able to get to. Cyril and I wanted to follow these up, so we have answered many of them below.
What is the estimated penetration index for NFV- and SDN-based services in the market?
: We turned to the webinar's 350+ attendees for their thoughts. We learned that of the 42 communications service providers (CSPs) that cast their vote, 43 percent will have launched services supported by NFV or SDN technology by the end of 2015.
Further, 40 CSPs responded to a separate question on their service assurance practices; 95 percent have considered service assurance as a critical success factor in the implementation of NFV- or SDN-based services.
These figures only further validate our thinking that the time to start planning the management of NFV and SDN is now, and operators must be ready to make the necessary fundamental changes to their OSS that will support a successful transition to NFV and SDN.
In the move from physical devices to virtual, will there be drastic changes in the way the FCAPS model operates?
: Based on feedback from TM Forum members, especially CSPs, we believe that realizing a truly virtualized infrastructure and enabling the end-to-end digital ecosystem will be disruptive to the current OSS/BSS infrastructure. For example, if you look at the use of physical location within billing systems or the procedures to address physical fault, there will be significant advances. You really begin to see this if you compare the way a virtualized data center operates with the current FCAPs model.
Network domains can be made of physical and virtual components, and the performance factors are different when identifying each KPI. With this in mind, what are the end-to-end KPIs for a network domain?
: At the end of the day, the end-to-end KPIs that are linked to customer service level agreements (SLAs) will not change. NFV and SDN will make those services more agile and will introduce virtualization of the last mile that will allow CSPs to offer a bundle of new, on-demand services, such as application visibility or managed security. But, through this transition, the parameters linked to an SLA are likely to remain the same.
When talking about end-to-end metrics, CSPs will need to determine in which business context they are being used. When talking about IP-VPN or Carrier Ethernet services, CSPs will have to measure the traditional end-to-end availability, response time and jitter, and will be able to leverage existing agents' technology offered by network function providers. But, as customers are able to order connectivity or increased bandwidth for a future date, CSPs will need to predict if they will be able to fulfill those future SLAs in order to confirm if a customer request should or should not be accepted.
: TM Forum is exploring this question of performance factors and KPIs within its Catalyst program. TM Forum agrees that the actual KPIs may not change, but the amount of physical information that is pushed through APIs may drastically change. This information will be collected from an orchestrator, rather than from a physical device, because so much of the network will become software-centric.
Will NFV/SDN bring changes to MTOSI standards? Is TM Forum working on defining interfaces with network devices for traditional services?
: This is subject to change. MTOSI is an interface designed specifically for legacy hardware and implementations, and will be best exploited as part of a hybrid network. TM Forum doesn't see that API actually being used much for virtualized infrastructure. Instead, it is on COTS hardware with software-based control mechanisms on top. We see that being more in the REST arena. It may have the same functionality as the implementation technologies that are emerging, which do seem to be far more focused on REST. This is something TM Forum has been exploring with its Catalyst program and with MEF. TM Forum wants to push the new virtualized capabilities of this trend into future implementations.
What kind of changes do you see in the traditional network planning and capacity management processes due to NFV and SDN?
: One of the most complex problems CSPs will face will be how to allow newly introduced orchestrators to assess virtual function load in a vendor-agnostic manner. Most network vendors are proposing orchestrators, and it is unlikely that one vendor will share the limit of their virtual functions with a competitor.
Would TM Forum recommend that CSPs automate their existing processes and systems on existing networks first, so that the hybrid network does not have the same problems in service agility? If not, what is the best way to address end-to-end service agility?
: This is a difficult question to answer generically and will vary based on the specific business scenario. The key factors to weigh, though, are the expected ROI for the automation of an existing process vs. how long you expect to exist in a hybrid environment. This question assumes that, for the hybrid environment to deliver agility, both environments need to be automated. It's more important to consider the cost of quickly deploying virtualization vs. the cost of automating dated processes that will one day be discontinued.
What is your opinion on how the performance assurance competitive landscapes will evolve as a result of NFV? Specifically, how will they evolve in terms of assuring the network and service performance domains
: The vendor landscape will change, in part, because most legacy performance assurance platforms will not make the cut in terms of managing dynamic services. For one, many solutions are lacking some real-time provisioning capabilities and the adaptability and flexibility required to manage these more sophisticated networks. On the network virtualization side, many vendors are coming from the IT domain, where they do not have a precise understanding of CSPs' constraints and challenges. CSPs' management infrastructure needs to scale to a greater level and needs to be integrated into the OSS ecosystem, which requires strong telecom expertise.
Interested in learning more? I invite you to check out these helpful resources:
- Recording of our webinar: NFV & SDN: The 5 Steps Your OSS Will Need to Take to Succeed
- Cyril's latest article in SDX Central: 5 Pragmatic Steps to Secure Your Transition to NFV/SDN-Based Services