logo_header
  • Topics
  • Research & Analysis
  • Features & Opinion
  • Webinars & Podcasts
  • Videos
  • Event videos

Is NFV in the infamous trough of disillusionment?

20 Jan 2017
Is NFV in the infamous trough of disillusionment?

Is NFV in the infamous trough of disillusionment?

Communications service providers appear to be losing confidence in their ability to meet planned timelines for completing virtualization of high-priority functions, according to Light Reading’s latest Future of Virtualization Indexes.

The semi-annual planning, deployment and spending indexes, which are part of Light Reading's Virtuapedia initiative, track operators’ progress in deploying virtualization. A total of 40 operators from around the globe, most of them Tier 1 providers, were surveyed in November.

The results, which were presented this week during a webinar, show that confidence in timetables for testing and deploying high-priority functions has declined significantly during the past year, with the percentage of respondents saying they are “extremely confident” in being able to meet them dropping from 22 percent in November 2015 to just 5 percent in November 2016.

High-priority functions include virtualized network functions (VNFs) with 39 percent of respondents saying they are the highest priority, followed by NFV and cloud services at 16 percent and enterprise and customer premises equipment functions at 13 percent. Less than 10 percent of respondents ranked other functions including operational and business support systems (OSS/BSS), video-specific applications, management and orchestration (MANO), storage, application enablement and compute as highest priority.

hr-webinar-2017-confidence
Why the loss of confidence?
“One would expect [confidence] figures to trend this way, but it does seem fairly dramatic to me when I look at these numbers,” said Roz Roseboro, Senior Analyst, Heavy Reading, who reported the results during the webinar. The reason for the loss of confidence likely lies in the difficulty of operationalizing virtualized functions.
“At the beginning of projects people are very optimistic, but now we’re starting to uncover the challenges,” Roseboro said.

A panel of suppliers who joined Roseboro on the webinar – Doug Tait, Director of Product Marketing, Oracle Communications; Lynn Comp, Senior Director, Market Development Organization, Network Platforms Group, Intel; and Marton Sabil, NFV Manager, Ericsson – agreed that management and orchestration (MANO) is the biggest challenge.
“Orchestration is hard,” Tait said. “There are too many standards, too much open source, too much going on. The bulk of our work is around orchestration.”

Comp agreed: “It’s not frictionless; it’s not push-button,” she said. “The expectation was that open source was going to make it that way and that’s not necessarily the case.”

Reasons for optimism


Tait said he’s encouraged, however, by the work going on in open source groups and at TM Forum to make virtual functions interoperable. Sabil was also optimistic.
“With Layer 3 standardization, we will have some good agreed standards coming in middle of this year that will remove some of the guesswork in the MANO environment,” he said. “There are many multivendor projects which is, of course, good – this is what we have been waiting for.”

Sabil pointed to Open Platform for NFV (OPNFV) and a recently announced agreement between Ericsson, Cisco, Huawei and Nokia to create the NFV Interoperability Testing Initiative in support of OpNFV as evidence that the industry is moving in the right direction.

In terms of actual deployment, the bulk of high-priority deployments are expected to take place in 2017, according to Light Reading’s survey. By the end of the year, 45 percent of respondents expect to have 20 percent of their high priority functions in live production networks, and 27 percent expect to have 50 percent of them deployed. But more than a third said it will take until 2019 or later to get all high priority functions into production, and 21 percent say they still don’t know when that will happen.

Spending and saving


The results of Light Reading’s spending index survey show that more than 75 percent of respondents believe CapEx budgets for virtualization will increase in 2017, while 20 percent said spending will be flat and 4 percent said it will drop.
“The highest portion of respondents believe spending on virtualization won’t level off or decline until 2020,” Roseboro said.

Operational expenditure (OpEx) results were almost unchanged in this survey, with most respondents still expecting to begin seeing lower OpEx in about 3 to 5 years. But the amount they expect to save (10 percent) is low from the panel’s perspective.

Both Comp and Sabil said some of their customers have seen as much as a 50 percent OpEx savings by implementing virtualization. As an example, Sabil pointed to a mobile virtual private network deployment.
“Typical customer onboarding was 20 days, plus 30 to 40 days to upgrade capacity,” he said. “By moving that whole network to the cloud and including automation, activation of new customers went down to 15 minutes and adding capacity went down to 1 day. These are enormous savings.”

Lifecycle management and metrics


One other interesting observation from the panel was that there is a great need for lifecycle management of virtualized functions and benchmarking metrics to help operators compare suppliers’ solutions and make sure they are interoperable.

TM Forum has been doing a lot of work in this area. AT&T, China Mobile, Orange, Telstra, Verizon and Vodafone together are sponsoring an ongoing Catalyst proof-of-concept project called Enabling Digital Marketplaces, which is working to automate the entire lifecycle of virtual functions. The most recent phase of the project used TM Forum Frameworx and TOSCA standards along with ETSI NFV concepts to develop a common metamodel.

The goal is to create an efficient marketplace – kind of a Network App Store – to automatically onboard independently sourced virtual functions and then use them to automatically provision, license, manage, assure and bill for services end to end across partners’ networks. This requires not only a metamodel but also metadata-driven metrics to automate assessment of each step of the cycle.

The Forum’s ZOOM team is working on the metrics framework with other SDOs and open-source groups including ETSI, NIST, QuEST and OpNFV, and the next phase of the Marketplace Catalyst will help with this work. We’ll be writing about this much more in the coming weeks, so watch this space.