Who Disaggregated My RAN? Part 8: Orchestrating the Open RAN Symphony

We left off in our series having looked at the virtualized and cloud-based open RAN. Now we move on to consider the orchestration part, deploying and managing these cloud-based implementations.  

 

Orchestration, or service management and orchestration (SMO), has been used in the enterprise world for some time to manage a number of virtual machines in a public cloud-based implementation in some form or another. Over time, the market has continued to develop specific orchestration products for a variety of vertical industries and has driven a number of innovations. 

 

SMOs provide lifecycle management of virtualized network functions (VNFs), such as creating new instances of an application, increasing or decreasing the number of instances to scale to meet the load, managing the multiple instances, and then shutting them down when there is no longer a need. 

 

 

For telecom, orchestration was originally described and established as a reference architecture in the early specification from the European Telecommunications Standards Institute (ETSI). In 2013, ETSI released the first NFV reference architecture implementations which established how telecom network functions can be implemented in a virtual machine environment and the technical requirements for such deployments. ETSI’s NFV reference architecture provided well defined interfaces for these network functions towards virtualization infrastructure and orchestrator as shown in Diagram #1.

 

As a result, an ecosystem began to flourish as OEMs started to implement this virtualization approach. Many operators started evaluating proof-of-concept implementations to understand VNF implementations and orchestration needs. Most of these projects and early implementations in LTE networks were targeted towards virtualizing EPC (core network) nodes. 

There have been a few open source projects to provide comprehensive orchestration capabilities for telco deployments. Two notable projects in this space are Open Source MANO and ONAP (Open Network Automation Platform). Of late, there has been widespread support for development and adoption for ONAP. ONAP was a culmination of two projects: OPEN-O (Linux Foundation) and AT&T's ECOMP (Enhanced Control, Orchestration, Management, and Policy) platform. Today ONAP has a strong development community supported by many operators and vendors. ONAP releases are now aligned to support Open RAN architecture and management/deployment needs.

Orchestration, O-RAN Style.

The O-RAN Alliance has focused from the start on cloudified implementations of Open RAN and related needs for management and automation. The Alliance’s Working Group 6 has been developing a reference architecture for a cloud platform architecture and the orchestration layer. O-RAN architecture envisions a uniform O1 interface for the management of RAN nodes. The O2 interface defines the architecture for managing the underlying infrastructure and orchestrating network functions as shown in the diagram below. A detailed view of how to handle various common scenarios is available in the O-RAN specifications as use case descriptions.

Diagram 2: O-RAN O-Cloud and SMO relationship

The other emerging trend in the RAN deployment is the need for colocating edge computing functions. Many companies, including Radisys, continue to experiment with this convergence of edge computing and RAN services. Companies providing commercial solutions of cloud infrastructure and orchestration, such as Ciena and VMware, are partnering with RAN vendors to provide deployable solutions to address use cases of edge computing, network slicing etc. 

Recently, Radisys and Ciena partnered on a joint demonstration which showcased Radisys’ disaggregated, virtualized and containerized RAN with O-RAN compliant O1 and pre-standard O2 interfaces leveraging Ciena’s Blue Planet’s intelligent automation software as O-RAN Service Management and Orchestration Framework to configure, deploy and scale network elements on demand and adapt to changing network conditions.

Orchestration and the Future of Network Operations

What are the benefits orchestration can bring to operators? In terms of actual savings and benefits to the operator, orchestration makes sure that only the required software instances are running, scaling up and down, health monitoring of network functions. The level of automation provides critical cost savings and the full benefit of a virtualized network. Operators are looking to bring in seamless, transparent experience, along with the increased capability that orchestration provides in order to manage the Open RAN network and allow it to run automatically. 

The closed loop of management, control and KPI reporting back to SMOs as envisioned in O-RAN architecture provides insights and immense opportunities to operators to continuously improve the network performance. With consistent and transparent KPIs collected from RAN nodes and UEs, there is a huge wealth of real time data available now for data mining. An operator can deploy the smarts of AI/ML and other techniques to derive policies and actions from the metrics streaming in from the network.

All of this points to a future where the automation is taken to its extreme levels. Operators are focusing on use cases that are far more advanced, as they consider how smart and intelligent the network can be, and how an orchestrator can play a role in making these advanced use cases a reality. The ability of operators to monetize various use cases and applications of 5G technology, orchestrated and managed by SMOs is going to be a game changer.

Stay tuned as we consider how an intelligent Open RAN can deliver one such use case – network slicing – in the next installment.