OSM Comes to NFV MANO With Code in Hand

View Original Article

On Wednesday at the Open Networking Summit (ONS), we heard from management and network orchestration (MANO) group Open-O. Thursday, the Open Source MANO (OSM) group got its turn, pitching a project that’s launching with code from three in-progress efforts.

Open-O and OSM are among a handful of open source NFV MANO efforts emerging for network functions virtualization (NFV). It’s hard to contrast the projects, considering they’re barely getting started, but OSM’s presentation put particular emphasis on developing a common information model for virtual network functions (VNFs).

Open-O mentioned this goal as well, but OSM’s representatives — from Telefónica, Canonical, and RIFT.io — emphasized the information model’s role in nurturing VNF development.

They compared an information model to the sheet music in a literal orchestra: It allows one centralized unit — an orchestration system or a symphony conductor — to communicate with a variety of players, any of which can be theoretically swapped out without disruption. (Don’t tell the violas.)

The VNFs are “the interesting and important part of this problem,” said John Zannos, vice president of the cloudplatform business at Canonical. “If we can make the platform and the orchestration uninteresting, we’d all be happy.”

VNF vendors are finding the greater NFV framework to be a challenge, said Lakshmi Sharma, senior vice president of strategy, architecture, and product management for RIFT.io.

“They are all attempting to build their own systems for orchestration and management. They have expertise in specialty areas; it would be wonderful if they can continue to spend their time on building those applications,” she said.

OSM’s place in ETSI NFV, noting the roles of Telefónica, RIFT.io, and Canonical. Source: OSM.

OSM’s place in ETSI NFV, noting the roles of Telefónica, RIFT.io, and Canonical. Source: OSM.

Interchangeability is at the heart of OSM’s work so far. The project’s early philosophy is to support multiple virtualized infrastructure managers (VIMs), the entities that create and manage virtualized elements such as computing and networking — even though Telefónica is contributing a VIM to the project.

In trying to come up with a framework across Layers 1-3, OSM’s backers also claim their project is looking at a broader scope than others are, at least so far.

Like other MANO efforts, OSM is targeting network creation as well as fulfillment and assurance, the latter two categories being the MANO pieces that would talk to a carrier’s OSS.

NFV MANO in Progress

OSM launched in February, assembling some in-progress pieces of MANO development — specifically, Telefónica’s OpenMANO, which launched last year, and the RIFT.ware platform from RIFT.io. Canonical contributed Ubuntu’s Juju charms, an established mechanism for modeling and configuring VNFs.

The joint effort emerged from talks among carriers in November at the SDN & OpenFlow World Congress in Düsseldorf, Germany. Most of the MANO attention at that time had gone toward VNF deployment and maintenance, OSM’s representatives said. A common data model was lacking, and some carriers believed the industry needed to reconcile the different orchestration systems that were emerging, said Gerardo García, a Telefónica R&D expert.

Speed was of the essence as well. “We need to have some kind of prescriptive decisions so we could have, very soon, an implementation that could be used by the VNF developers,” García said.

To jump-start their efforts, OSM pushed to get a demo prepared for Mobile World Congress last month, where they showed off a model for a Layer 3 VPN.

The approach is meant to be modular. Search orchestration and resource orchestration can be decoupled, for example. “If someone feels that service orchestration can be handled in a traditional way, with OSS, it can be done,” García said.