With the imminent success of cloud operations and virtualization in the data centre and application world, now the focus has shifted to the networks to be virtualized and to leverage cloud infrastructure as common base. As the acceptance of 'Softwarisation' of Network Gears or function gradually sets in, virtualization of network features and gears is on rise. However, the big promise of network virtualization with faster service rollout, dynamic Service life cycle management, and improved Operational efficiencies, can only be delivered with right set of tools and techniques for creating & managing new services, managing the 'new normal' network comprising of PNFs and VNFs, and ensuring that the right instruments are brought on board to perform the orchestration job for architecting the services and eventually network.
In simple words, Orchestration is an automated, well-coordinated managed series of complex tasks. Ideally, Orchestration should encompass the inherent intelligence to provide autonomous control.
Fig. 2 : Cloud of tasks for Telco network management, operations and orchestration
Hybrid Resource Management
While a majority of Service providers began the network virtualization journey by piloting VNFs in 'contained' business domains, their real production challenge would still be to handle hybrid resources (Physical and Virtual), multi-domain technologies starting from cpe, access, transport, edge, core, data centre, and backend applications across multiple vendors. The challenge arises with creation to manage, monitor, and fine tune the performance of the services/network in a hybrid environment. Each domain, technology, and vendor comes with its own set of supported Orchestrators and SDN Controllers. This essentially creates a cocktail of orchestrators and controllers. This cocktail could easily land in another vendor-lock in scenario for the operator, unless a pragmatic approach is applied for orchestrating the complex system.
Operator's Choices on Orchestration Approach
Clearly, there is enough ground for a special orchestrator system which can handle the multiple layers of infrastructure, network functions, and services, along with the operator's backend system. There are different ways how large and niche players and standardisation bodies are driving their PoCs to address this space. In the future, there would be different players to compete for this space and a larger pie from the operator's spend. Let us review what technical approaches do the operators have for multi-domain, multi-vendor orchestrators
- Single Layer Approach and
- Multi-layer Approach
Orchestrator using Single Layer Approach
A Single Layer Approach is a solution with an end-to-end Service Delivery and life cycle Orchestration. This eliminates the management silos of the distinguished technologies and domains by integrating with domain specific controllers and orchestrators. This provides an interface to work with B/OSS Operators backend applications. Product like by Blue Planet from Ciena/Cyan are taking this path and chasing the space.
Orchestrator using Multi Layer Approach
A Multi Layer Approach is a solution which breaks the domain silos by unifying the resources across domains and technologies, including physical and virtual. This approach sets a path for layers of orchestrator at function level. Layered Stack of orchestrator starts at basic infrastructure (hardware and software level for compute/Storage aka cloud) as infrastructure orchestrators, Network function layer (including VNFs and PNFs and its management) as Multi-domain NFVO, and last but not the least, eventually at Service Orchestrators at business layer, which will eventually lynchpin to customer self-service portals. UNIFY is an EU-funded FP7 project aiming to unify the cloud and carrier network resources in a common orchestration framework, demonstrating the multi-layer orchestration approach with SFC deployment with multi-domain.
Operator's Interim Solutions
With the evolution of orchestration engineering, clear winners and solution for operators' Hybrid Resource Management are yet to emerge. However some of the solutions that the operators can implement at this stage are:
- To partner with an Orchestrator player spanning across domains
- Choose a solution which extends/provides OpenAPI/ Standard template to interwork with backend applications, VNFs and 3rd party controllers/orchestrators through north bound , south bound and east-west interfaces.
- Add the plug-ins for APIs, work on Data lake as needed
- Build a custom higher order orchestrator system spanning across domains and resources (physical and virtual).
- Experiment with open sources and create the custom roadmap for engineering integrating with domain controllers and orchestrators.
- Experiments and learn the technology as it matures and evolves with deployment.
- Define the services and resource map and create an orchestration algorithm
While Network orchestration and management systems (NOMS) revenue worldwide is forecast to grow from USD4.7 billion in 2014 to USD5.3 billion in 2019, at a 2.6% CAGR, after a 2% growth in 2014. - See more at:
http://www.analysysmason.com/Research/Content/Reports/network-orchestration-forecast-Nov2015-RMA07/#sthash.gRc9SEom.dpuf. Eventually, Operator's required Orchestration Engineering will evolve, approaches shall refine and clear winners are still to emerge until then it remains as Cocktail of Orchestrators.