In this second blog post about the TM Forum report on Orchestration, we’re going to look closely at the definition of orchestration as it pertains to today’s virtualized mobile networks.
“In general, the definition of network orchestration is too narrow and too specific to VNF lifecycle management. Operators have backed away from talking about OSS recently, fearing it sounds ‘retro’, but it is clear we still need a top-level layer of intelligence to manage end-to-end services, which is what OSS has traditionally done.” – Ron Parker, Chief Architect, Affirmed Networks
Defining Network Orchestration
The basis of the TM Forum report was a broad definition of network orchestration as “end-to-end service management through zero-touch (automated) provisioning, configuration, and assurance.” After speaking with contributors to the report, what was clear is that orchestration, as it’s being implemented in live networks, is happening at multiple levels or layers, influencing virtualized and physical functions, including OSS, BSS, and NMS.
Generally speaking, there are four places in an operator’s environment where some kind of orchestration can take place.
Why Do We Need Automation?
“What you’re setting out to do is abstract the complexity and drive modularity. “If you imagine a future network state where everything is virtualized – all software-defined networks – every customer might have a completely different set of virtualized functions. In a traditional approach, you’d never get over that complexity – if something were to break, you’d never be able to fix it.” – Dr. Lester Thomas, Chief Systems Architect, Vodafone Group.
Service providers, like Vodafone, AT&T, and others, are proposing is that there will be multiple platforms within the network, each abstracting some of the complexity. As examples, Dr. Thomas points to OpenFlow abstracting the complexity of an individual router, while NETCONF and YANG abstract SDN controllers. The overarching point of abstracting at a higher level is to simplify the network orchestration and utilize “intent” to manage the policies.
We’ll get to that topic in a future blog post.