It’s ridiculously early in the existences of Open-O and Open Source MANO (OSM), two open source NFV management and network orchestration (MANO) efforts that emerged at almost the same time this year. But it’s not too early to spot differences between the two.
Specifically, each group hopes to solve different problems. OSM will focus on network service orchestration, which fits on the right-hand (MANO) side of the ETSI NFV MANO diagram (see below). Open-O will expand the scope of its work beyond MANO to include orchestration over the entire network.
OSM announced itself in February, and it’s based on the OpenMANO work begun by Telefónica about a year ago, with initial code also contributed by Rift.io and Canonical. The group is hosted by the European Telecommunications Standards Institute (ETSI). In addition to Telefónica, other initial service provider members include BT, Telekom Austria, Telenor, Korea Telecom, SK Telecom, and Sprint.
Open-O announced itself in March, and it’s hosted by the Linux Foundation. It’s starting life with a heavy Chinese influence, with founders including service providers China Mobile and China Telecom. And Huawei has pledged $30 million for the project over the next three years.
Although OSM is curated by ETSI and Open-O is not, the ETSI NFV MANO diagram is useful to explain what problems each group hopes to solve.
“There’s a disconnect between current architectures and stacks,” says Vincent Spinelli, SVP of global sales and marketing for Rift.io. At the virtualized infrastructure manager (VIM) layer, many service providers are using OpenStack, and for OSM, Telefónica is contributing an OpenStack VIM to the project.
“Then you have the SDN orchestrators controlling the physical layers,” he says. Those are different from the end-to-end service orchestrators — which, on the ETSI NFV MANO diagram, sit across from the VIM at the bottom rung of the OSS/BSS chain.
“What’s missing is above the VIM; we call it network service orchestration,” says Spinelli.
One thing OSM will be focusing on is development of a common information model to describe network services, whether based on physical functions or virtual network functions (VNFs). OSM will use model-driven languages such as YANG and TOSCA as a way to translate the description of network functions into viable services.
“One of the key attributes of this level of orchestration is focusing on actual services that service providers can deploy to either save money or make money,” says Spinelli. “OSM is very practical in providing that kind of focus.”
Open-O’s scope seems bigger, at least at this stage of the game.
“ETSI has been more VNF-centric, not a network-centric view,” says Marc Cohn, VP of network strategy at The Linux Foundation. “We want to address end-to-end service orchestration over any network, not just over an SDN network. We’re not just addressing the MANO problem. We’re also addressing the connectivity element.”
But Open-O’s work will be functionally compliant with the ETSI NFV MANO.
“Where it goes beyond that is where we get into SDN orchestration,” says Cohn. “That’s the primary difference between the two approaches. SDN-O [Open-O’s SDN orchestration element] is how we’re going to bring connectivity services into the overall service orchestration over multiple SDN controllers and network and element management systems.
“SDN-O focuses on how to interconnect the VNFs and provide an end-to-end service. A global service orchestrator provides end-to-end orchestration across the NFV and network (including SDN) domains.”
Open-O also plans to address NFV MANO with a modular approach that integrates with multiple VNF managers and VIMs.
Other NFV MANO Work
Since OSM and Open-O announced themselves, AT&T has come out with its Enhanced Control, Orchestration, Management, and Policy (ECOMP) platform, and Verizon made its SDN/NFV Reference Architecture public.
Spinelli says with AT&T, OSM sees the work it’s doing with MANO as a “sort of plug-in to ECOMP.” And similarly, he says Verizon has a need for network resource orchestration as opposed to end-to-end service orchestration.
Cohn says Open-O is similar to ECOMP and Verizon’s NFV reference architecture, but those are Verizon and AT&T developments and not open-sourced (at least not yet in the case of AT&T).
“Open-O represents what large operators are thinking about in terms of delivering end-to-end services — MANO, but also a bigger picture,” says Cohn. “The bigger picture is orchestrating any service over any network.”
Open-O is hosted by the Linux Foundation as a pure open source project, while OSM is being overseen by ETSI, a standards organization. Both hosting organizations have experience bringing together operators and vendors to drive use cases.
Ultimately, there could be room for both: OSM could conduct standards work within the ETSI NFV MANO diagram, and Open-O could expand upon ETSI’s NFV framework.
Cohn says, “We don’t think about standards at all. We comply and consider, but at the end of day we’re about software.”
There have been some raised eyebrows about Open-O’s heavy Chinese influence, especially the fact that Huawei immediately promised $30 million for the project. However, it should be noted that Open-O has non-Chinese members as well, including Brocade, Ericsson, F5, Intel, Red Hat, and Riverbed.
OSM has a jump on Open-O in that Telefónica has been working on its OpenMANO project for more than a year, and its code has now been folded into OSM. Francisco-Javier Ramón, head of the Telefónica NFV Reference Lab, said at the Layer123 NFV World Congress last month that OSM already has more than 25 companies participating, which include both ETSI and non-ETSI members.
Open-O is just getting off the ground. It plans to announce its seed code contributions in mid-summer, and it’s working on getting its governance model in place.