MEF has developed a reference implementation of lifecycle services orchestration (LSO) Analytics using Platform for Network Data Analytics (PNDA). The organization promised a demonstration at some unspecified date.
Many service providers that operate disparate networks find software-defined networking (SDN) and network functions virtualization (NFV) attractive because virtualizing should enable them to manage their multiple networks as if they were a single unified system. That requires some overarching management system – an orchestrator. And if the orchestrator is to function effectively, it requires a constant flow of network operations data, along with analytics tools to make use of that data.
That has led to the development of LSO Analytics. Some MEF members have proprietary LSO products while others share open source versions. MEF is coordinating the concurrent development of orchestration tools and both private and open source analytics tools through its OpenLSO analytics initiative.
MEF vowed to demonstrate how multiple open source and proprietary LSO products integrate with PNDA. The proposed reference implementation will illustrate how big data analytics (BDA) can be used within the MEF LSO Framework to intelligently automate configuration and the allocation of network resources in real time.
As of last summer, PNDA is a Linux Foundation project. The first demonstration of PNDA occurred in November and provided performance analytics for services provisioned by LSO. That demo illustrated how PNDA makes it possible to bring together multiple datasets, which previously would have been in separate silos, add service context from LSO, then enable rapid prototyping and development of analytics applications. The Linux Foundation has been demonstrating PNDA at the Open Networking Summit (ONS) in Santa Clara, California, this week.
Going forward, MEF’s OpenLSO Analytics project will focus on demonstrating analytics as part of Third Network services user stories, enabling analytics to span multiple service providers, and feed requirements into MEF specifications work. PNDA can be leveraged to provide operational and business insights and customer analytics outcomes, MEF said.
MEF and PNDA experts expect to have a progress report on OpenLSO Analytics at the Euro17 LSO Hackathon held during the 2Q17 MEF Members Meeting in Frankfurt, Germany, April 24 through 27.
The basic problem LSO Analytics aims to solve is that any given service provider is likely to have its network data distributed across multiple isolated silos, explained Dan Pitt, senior vice president of MEF.
An operator with ordinary packet switching networks would need to have information about load, buffer, flows, latency, and latency variation. But there would be additional information to tap at the physical layer that might include optical signal to noise ratio, bandwidth per wavelength, and other measures. If the company also operates a wireless network, that represents more variables. Operations in the cloud represent another set of data points, Pitt explained.
Companies need to synthesize information about the state of their networks in order to meet their objectives and business priorities, Pitt told SDxCentral. SDN requires an open platform for getting and assessing that raw data to turn it into useful analytical feedback. That’s what MEF is trying to facilitate with its programs, including OpenLSO Analytics and OpenLSO’s interoperability with PNDA.
“The Linux Foundation is pleased to share a common vision with MEF on the critical role that open source big data analytics can play in accelerating the transition to more agile, assured, and orchestrated services,” said Marc Cohn, vice president of network strategy for The Linux Foundation in a statement. “PNDA’s open framework aligns with MEF’s role in defining APIs, and its options for data collection are on the forefront of the leading thinkers in the field.”