As the demand for true network virtualisation is about to be met, a key issue is how CSPs can deliver the best SDN experience. The importance of capacity planning, management and orchestration is now being recognised in the journey towards end-to-end virtualisation, writes Dr Jay Perrett
Similarly to software defined networking (SDN), network functions virtualisation (NFV) is one of the most established components of network virtualisation. Network virtualisation provides CSPs with the opportunity to derive more value from operational and planned network and data centre assets. This is achieved when deployment and redeployment of network assets and functionality to meet demand takes place through machine rather than human control. Virtualisation has become essential if new and emerging business models, which are more data intensive than ever before, are to be effectively supported.
Two different industries, each with a language of their own, are facing the challenges of network virtualisation from two different angles. The first is the data centre community, which has been moving toward reusable assets in the data centre to provide virtual machines to users that can rapidly be configured. The second is the CSPs, which need to provide new services and functions dynamically and to an aggressive timescale. Both have the same objective: to be able to deliver a virtual network that can be configured and reconfigured dynamically without constraints of geography or hardware.
NFV and SDN provide a unique opportunity to bring IT and telecoms capacity planning together. Historically, IT and telecoms capacity planning functions have existed in very separate silos, using their own processes and methodologies. With NFV, the consolidation of IT and telecoms capacity management functions is needed to match overall capacity and business demands, as well as to be able to obtain true end-to-end visibility of the networking real estate.
Capacity planning
From a capacity management perspective, if the transition to a virtualised network is to be realized, three key requirements must be addressed. First, a generic representation of the network must be developed that can be instantiated for the technology mix under consideration or operation. A data driven representation of capacity will also need to be created, so a clear definition can be described in reusable templates. Finally, there must be a flexible way to create a process flow.
Capacity management is only one aspect of a managed virtual network, but it is an important requirement in the orchestrator layer. Adaptability is the key facet of a virtualised network. What this means from a capacity management perspective is the ability to respond to changes in network infrastructure. To remain competitive, the network needs to be abstracted from the technology – the network is simply a vehicle for the support of services and applications and thus revenue. With this approach it becomes simpler to configure and rapidly reconfigure the network model simply by changing the data required to model it. This delivers a software defined network.
The second key requirement, that a data driven definition of capacity facilitating a more detailed level of control, is required to effectively manage capacity on a rapidly reconfiguring network. The capacity definition and capacity requirements need to be data driven also. This has been achieved in the past through traditional intelligent networks, where service function was built from smaller building blocks, but the equipment those services ran on was still bespoke. The definition of capacity also needs to be abstracted from the equipment, which is essentially what NFV enables by abstracting the network function from the physical equipment. Once capacity has been abstracted, services or applications can then utilise the capacity they require without being constrained by how./p>
Orchestration
In a network the orchestration layer is performing the role of network operating system. It is responsible for separating the infrastructure from the applications by providing a standard application programming interface (API) to the applications in one direction and to the network in the other.
Vendors who provide management and orchestrator functionality for services and applications will need to model networks in a different way. Specifically they will need to consider a data driven definition of both the network and capacity. The former is really an SDN enabler. Once networks can be defined simply by data, any network architecture can be modelled or tweaked in real-time. How the network functionality changes is a defined by the second requirement, the NFV enabler.
Capacity management and orchestration are front and centre of the evolution to a true service enabled virtual network constructed on the principles of NFV and SDN. Network operators need to effectively orchestrate capacity to derive the best possible value from infrastructure assets to deliver the most effective capacity management. Central to this objective is effectively understanding the current and future demands on a network and enabling it to optimise itself in real time to deliver to those demands. If CSPs are to remain competitive in an environment with eroding margins and rising service quality expectations, the virtualised network route with a SDN supported by NFV really is the only option. Whilst the technology and business drivers to achieve this already exist, two very different network communities need to converge on a single view of the virtualised network.
A new approach is needed. An approach that thinks of networks, capacities and consumption as enablers to the provision of services and applications to a population of consumers who rarely consider the network between them and their content and who also do not expect to pay much for it. In this environment, understanding the cost of delivering capacity, maximising capacity from current network assets and having a clear and transparent return on investment route from network investment are essential elements to delivering profitability.
