There are numerous attempts afoot to wrestle NFV into a more manageable and workable approach to virtualization.
Last week at the Open Networking Summit, some of the carrier members of a new effort around simplifying network functions virtualization infrastructure (NFVi) presented their approach on a panel.
The group, which is called Common NFVi Telco Task Force, is comprised of AT&T, Bell Canada, China Mobile, Deutsche Telekom, Jio, Orange, SK Telecom, Telstra, Verizon and Vodafone.
Currently, there are too many types of NFVi floating around, which means virtual network functions (VNFs) vendors need to create multiple versions of their VNFs to work with the different flavors of NFVi. The Common NFVi Telco Task Force is taking aim at reducing the number of NFVi implementations down to three or four versions, according to AT&T's Amy Wheelus, vice president of network cloud.
"What we're trying to do is get to a finite number to start with," Wheelus said. "We want to start with what's the first one, then let's add the second one, then the third one, but we don't want it to be 10 versions of NFVi. We want to make it a finite number."
The task force is working hand in hand with LF Networking's testing and compliance program, OPNFV Verification Program (OVP), for VNFs, which was announced last week at ONS.
RELATED: LF Networking expands compliance program to include VNFs
The goal of the OVP testing program is to ensure commercial NFV and VNF products are ready for the market in a timely manner. The NFVi Task Force would take those certified VNFs into the versions of NFVi that its members pick as the best options.
"We're not trying to create another organization," Wheelus said. "We're just trying to put legs to what's there. We, as service providers, would get that VNF that's been stamped with the OPNFV certification. So, we know that it's already been through this testing that we've contributed our requirements to, that we've contributed test cases to.
"The piece that that is taking it a step further is that we want the service providers to really lean in and help contribute test cases, so that we create this automated system of testing. We can run that test case in NFVi just to make sure everything works. We don't have to spend six months testing a VNF. It's already been done, it's already been certified. We just run it as a validation test."
Wheelus said the task force also could pull from other open source communities, such as using Akraino's network cloud blueprints.
In the current world of VNFs and NFV, operators and vendors are not getting economies of scale due to the time it takes for service providers to make sure a VNF runs on their versions of NFVi, and the vendors needing to create multiple VNFs for each NFVi.
"What we realized is that on the NFVi side we've been a bit limited to-date about what we test," said Heather Kirksey, vice president, community and ecosystem development the Linux Foundation. "Because there are so many flavors of NFVi out there you kind of end up going a bit to the lowest common denominator. The operators are feeling the pain from having made very different choices early on. They're not getting that sort of ecosystem level economies of scale. So the initiative is sort of the operator saying 'Okay, to enable this program vision, to enable us, even more than this program, to get those economies of scale we as operators need to agree to more commonality."
Wheelus and Kirksey said the Common NFVi Telco Task Force is looking at work that GSMA, Orange and Vodafone have done in regard to writing standards around common definitions for NFVi. Wheelus said AT&T, Verizon and other members of the task force have started to participate GSMA's effort.
"This will kind of marry the standards in the open source world together on this in that you've got a standard that will be a document that's a living document that will get submitted into GSMA that we will use that to create the reference architectures that will then be built in the OPNFV world," Wheelus said. "The beauty of setting this up within OPNFV is that feedback loop so that the providers can continue to get feedback because the infrastructure is going to continue to evolve.
"Today, we're using VMs tomorrow, we'll use containers. Today, there are very few GPUs (graphics processing units) deployed, but tomorrow there'll be GPUs. So, you've got to continue to evolve this. If we can keep that focus within OPNFV where we're doing it once, and we're not all doing it individually, it will help the entire industry."
Wheelus said infrastructure is not where telcos should be competing against each other. Telcos can work together on building the testing and compliance platform and the NFVi ecosystem while competing on the service level.
She also said there's a proliferation of projects within the Linux Foundation and across the industry, but telcos and vendors need to avoid competing against each other across all of those projects.
"We're trying to consolidate and work together to go drive this," Wheelus said. "We've talked about our roadmap, and we're going to talk some more about it, but we would like to have at least one of these reference architectures stood up, and the testing process in place by the end of the year. We're kind of targeting the September ONS In Europe to come together and really put more structure around those next steps. We'll spend the time between now and then to get at least one common architecture in blueprint format."