BARCELONA, Spain – The open source ecosystem made another significant push into the traditional telecom market with the launch of a new Cloud native Network Function (CNF) Testbed that could speed the rise of Kubernetes as a test platform. That rise would come at the expense of OpenStack.
The test bed was developed as a partnership between the Cloud Native Computing Foundation (CNCF) and the Linux Foundation’s LF Networking (LFN) group. Its purpose is to show the ability to run the same networking code running as virtual network functions (VNFs) on OpenStack and as CNF’s on Kubernetes.
CNCF Executive Director Dan Kohn explained that the CNF path provides performance improvements due to a lower virtualization overhead.
For telecom-focused vendors and operators, the test bed allows for a consistent and reliable method to test network functions from ONAP or an organization’s own networking code as VNFs and CNFs. Those organizations can then compare the performance and resiliency between running on Kubernetes and OpenStack using the same underlying hardware.
“We don’t want them to take our word for it,” Kohn said. “We want them to try it and see for themselves.”
The CNF Testbed uses open source VNFs from ONAP’s virtual customer premise equipment (vCPE) use case. It also repackages the code as containers to be CNFs.
The Testbed taps into the Community Infrastructure Lab that uses “credits” provided by bare-metal hosting company Packet. Kohn added the Kubernetes project is also running 10,000 continuous integration (CI) jobs each weekday using funds from Google’s $9 million commitment when it fully spun off the project into CNCF last year.
Kohn explained that the testing can be conducted at an organization’s own environment, which should alleviate any security concerns over proprietary information or code used in the test. However, he stated that the test bed does ask for the results of those tests to be shared with the test bed in order to fine tune performance.
Kohn also noted that one of the benefits in moving down this road is that it focuses an operator or vendor on a “higher development philosophy.”
“Once you move to containers, which is CI, it should change the development approach,” Kohn said. “You can now do updates daily instead of monthly and fix bugs faster.”
This could still be a big ask for the telecom space, which operates in a legacy mindset encumbered by legacy hardware and regulatory challenges. In fact, in speaking with a handful of vendors during the MWC Barcelona event on how software was impacting their CI plans, many were excited by the thought of moving from a yearly update cycle to something that was quarterly or even monthly. Daily? Surely you jest?
Where’s This Going?
Kohn said this move also sets up a future where “Kubernetes will be the universal platform that can abstract the bare metal and any cloud.”
“And on top, all functions can be cloud native functions,” Kohn said. “This allows for an efficiency advantage in that all OSS and BSS can also run on the same cluster that is running the networking code.”
One example of the telecom industry’s growing confidence in Kubernetes is the Airship Project launched last year by AT&T, SK Telecom, Intel, and the OpenStack Foundation. The initial focus of that project is the implementation of a declarative platform to introduce OpenStack on Kubernetes (OOK) and the lifecycle management of the resulting cloud.
AT&T, specifically, is using a network platform from Mirantis that has OpenStack running on top of Kubernetes. Mirantis CEO and co-founder Adrian Ione explained that its work showed that the vendor was able to bind the Kubernetes substrate to bare metal and that it could get that substrate to work at scale.
Ionel said that Mirantis expects the platform to run a few thousand nodes this year, and then scale to 10,000 nodes over the next three years, and more than 20,000 nodes “in the years to come.”
“This is really about Kubernetes taking a prime role in the future infrastructure of a gigantic carrier,” Ionel said. “The scale of this is really staggering.”