Openet has built the first worldwide production deployment of specialized Telco-grade virtualized cloud offering for low latency PCC applications. The solution leverages an infrastructure as a service (IaaS) architecture to cover hundreds of hypervisors in 11 data centers tuned and optimized to support 100 million customers and over 1 million transactions per second across the North American continent.
Aims of NFV
Many of the main aims of NFV are based on the benefits that have been reaped in other areas using cloud and virtualization technology. These objectives can be grouped into three broad groups, cost, management of the deployed network functions and business agility.
Costs can be reduced as virtualized functions share capacity on standard commercial off the shelf (COTS) servers. Capital expenditure is reduced as lower cost servers are deployed whilst operational expenditure falls due to lower energy bills, and a higher ratio of network administrators to servers.
Simplified management and deployment of new server functions together with rapid scalability is simplified through use of virtualization and management/orchestration. Operational efficiency is improved in terms of easier management of the network functions such as managing scaling, failover and upgrades, as well as simpler hardware management due to consolidated hardware. Similarly, testing is more efficient as all functions run on the same infrastructure.
Business agility is enhanced as test cycles consume less time and fewer resources and service introduction can be rapidly scaled up and down.
Furthermore, the opportunity for innovation is increased as the barrier to deploy new network functions is low, meaning that network operators can experiment with new functionality and deployment models. NFV also opens up the potential for operators to share functionality, or to deploy network functions on each other’s infrastructure.
There are a number of challenges to achieving the aims of NFV across the entire network.
The use of industry standard hardware and virtualization will have an impact on efficiency. While this is not a significant barrier to PCC products (minimal CPU overhead due to virtualization), it will have greater impacts on data plane elements where custom ASICs are commonly used rather than off the shelf, general purpose CPUs.
Migration paths from current networks to virtualized networks may be complex. Virtualized functions will have to work in a hybrid model and interoperate with physical network appliances which are all managed by the network operator.
To achieve the true aims of NFV, the orchestration layer must be capable of managing all network functions from a multitude of vendors. A disparate community will need to produce common management and orchestration layers.
Automation and Scalability will be maximised when all network functions are virtualised. As NFV is expected to occur in a piecemeal manner, the full benefits will not be realised until all network functions are compatible.
The potential benefits far outweigh the obstacles in the path of NFV adoption. Many of the challenges listed above are being addressed currently by the work in the NFV ISG. Indeed, Openet itself as a pioneer of virtualization has overcome many of these challenges in large scale deployments.