Virtualization originated as a way for communication service providers (CSPs) to apply IT virtualization technologies to the network to help accelerate the deployment of new network services and remove the constraints of hardware-based appliances.
The advent of these new virtualization technologies, such as network function virtualization (NFV) and software-defined networking (SDN), have introduced a new layer in the network architecture: the orchestration layer. As defined by the European Telecommunications Standards Institute (ETSI), the NFV Orchestrator provides management of NFV services, which is responsible for on-boarding of new network services (NS) and virtual network functions (VNFs).
However, virtualized network functions will not completely replace the legacy network deployed in specialized hardware, and with that, CSPs will operate hybrid networks for the foreseeable future. The orchestration layer should expand from the data center all the way to the access layer, covering both physical and virtual network components. This orchestration layer must be integrated with legacy systems like OSS/BSS and EMS/NMS as well as with new elements such as network controllers or cloud managers (VIM, VNF Managers, etc.).
Opportunities and benefits
SDN and NFV are two independent, albeit closely-linked, technologies that have transformative potential to improve the customer experience and reduce costs. In conjunction with these technologies, the orchestration layer enables CSPs to make greater use of automated processes in troubleshooting, fulfillment and assurance. For example, the orchestration layer can help CSPs effectively become more agile by enabling the following functions:
· Reducing the time to introduce new services from months to days
· Creating on-the-fly network functions and moving them dynamically to a different hardware platform
· Auto-scaling network functions to optimize the hardware resources needed
· Auto-scaling network bandwidth and creating on-the-fly network services.
Realizing these benefits is predicated on virtualization and orchestration being done right—with automation and agility as its cornerstones. A new breed of analytics must be used to drive smarter orchestration for CSPs.
Likewise, the success of these virtualization initiatives will depend on the CSP’s ability to deliver high service quality and experience to expectant customers. Sophisticated, machine-driven analytics is essential to quickly detecting and redressing service disruptions in this complex ecosystem.
Analytics is the key requirement for automated orchestration and service assurance
With virtualization, CSPs enjoy a new level of flexibility that comes with the ability to instantiate network functions and services anywhere. Additionally, different functions and services can now share the same physical resources. This provides economies of scale. By separating the application layer from the hardware layer, multiple systems can be maintained via a single orchestration software. CSPs no longer need to have a separate management system for each piece of hardware.
However there are downsides also. Proprietary management systems of specialized hardware can easily monitor the performance of both the application and the hardware – as they interact together – and determine if a slow down or problem in one is affecting the other. With virtualization, the control plane is separated from the physical plane and the orchestration layer cannot see if what is happening at the physical layer is affecting the performance in the application layer. The complexity of identifying and isolating issues when they appear, as well as pinpointing their root cause and determining the true impact to customers, increases with virtualization. Virtualization creates the need for much more sophisticated assurance capabilities that go beyond the traditional network monitoring to include data center and network analysis capabilities – encompassing both physical and virtual resources in real time to bridge the gap.
The current OSS layer will also need to adjust to a new model in which control and service functions are instantiated dynamically based on resource demands. In addition, the virtual network functions (VNFs) and the links used to connect services must move dynamically based on utilization, policy, and business logic rules. As VNFs are added, moved, and de-commissioned, the monitoring points must also be created, moved, and retired at the same rate. The service assurance system should have a unified, end-to-end view of network topology (both physical and virtual) and operate in real time to detect issues and recommend solutions.
Orchestration adds a new layer of complexity to coordinate, monitor and manage all the resources involved in delivering services. Orchestration must be supported by an analytics engine that can collect information in real-time from all different parts of the network, both physical and virtual, and correlate the data with other OSS/BSS systems as well as customer experience markers to detect anomalies, predict customer experience issues and prescribe the solution to the orchestration layer. This will enable dynamic resolution of network problems and troubleshooting as well as automatic service assurance, which in turn will improve the customer experience.
Only then will the true promise of virtualization in CSP networks become a reality.
Part 2: Next week we’ll review a checklist of what’s needed to manage new virtualized solutions and give an example of how analytics are filling the technology gap that exists today.