NFV and SDN are tremendous opportunity for telecommunication operators to transform their networks and clouds in low cost and elastically scaling environments. But to fully achieve these promises, Operators have to tight NFV with a single view of their subscribers (SDM) and with Analytics
It is well-known that Telecommunications Operators are facing hard challenges these days, due to the combination of Over-The-Top services competition, exponential growth of mobile traffic and increase customer’s expectations in terms of quality of services. The result is a hyper-competitive market, where the customer experience level is the key to succeed. In this scenario, Customer experience is no longer mere network function, but it involves the whole operator, from network to business, from Customer care to marketing.
A superior Customer Experience Management (CEM) implies to cover at least 3 main aspects:
- It requires a more personalized marketing system, capable of operating on customer’s preferences, by proposing personalized offers and products.
- It requires a very flexible and interactive Customer Care capable of answering to any question/request from the customer.
- it requires a very flexible network and services management, where resources can be dynamically allocated based on customer’s profile but also on location, time-of-the-day, bandwidth available, device type, etc.
While all these 3 aspect are equally important to manage the CEM, the third one got the main attention from Telco due its impact on the investment side. Network virtualization, indeed, allows Telco to strongly reduce network investments to increase traffic bandwidth.
Software Define Network (SDN) and Network Functions virtualization (NFV) play a crucial role in creating a network and services environment that at the same time are extremely flexible and cost efficient. NFV and SDN introduce into the network a cloud-Like model – standard hardware, separation of logical and control plane, centrality of software for management, etc. – allowing operators to create flexible easy and cost effective networks.
Indeed, the main scope of NFV and SDN is to implement each network function as a pure software running on industry-standard hardware. In other words, in a NFV/SDN environment the hardware and software parts of any network elements are decoupled and they can run on different hardware. This break of the traditional network model, brings higher levels of modularity through virtualization of all software functions and introduce the benefit of cloud to the network : resource optimization, scale-in/out flexibility and automation. Beyond that, it also allows functional separation – logical and physical – between the data flow plane and the traffic control plane. The unbunderling of control plane enables new level of functional automation and flexibility.
Virtualization of the control plane
Next generation telco network environment requires virtualization strategy to go beyond the Data plane and cover the Control Plane. With the virtualization of Control Plane functions such as PCRF (Charging Rules Functions) , HSS (Home Subscriber Server), SDM (Subscriber Data Management), OCS (On-line Charging System) or MME ( Mobility Management Entities) – networks will become more flexibile and cost-effectively than ever.
With this evolving model of Network control plane, Policy and Control functions become instrumental to more activities of service operations:
- Quality of service (QoS) and experience (QoE) – Dynamic allocation of network and services resources, using customer’s profile information that is converged from many sources: billing info, location, status of the networks and applications, device type, service entitlement, and more
- Real-time usage monitoring – including bandwidth control and real-time network resource allocation
- Secure access, controlling if customer (in person or through M2M connectivity) is entitled to access specific resources, services and applications in any given time and location and devices
While Control Plane virtualization is vital in 4G and LTE environments, it also plays an important role when considering existing 3G/2G infrastructures. An integrated 4G and 3G/2G virtualized control plane allows Operators to consolidate their subscriber base in a common repository (UDR) simplifying the subscriber management across different RAN technologies.
Closed loop decision systems
In order to fully deploy the potential of NFV/SDN, the combination of virtual data plane and control plane is still not enough . Control plane must be connected to real-
time Analytics engine to create more powerful and advanced close loop decision systems. Indeed, most of today`s policy systems are based on static condition-action rules, where known conditions determine what and when actions should be performed. Decisions rules are limited to network segments, and cannot consider a whole network status or learn from historical situation .
For preventing network’s issues (congestions, offload, social events peak traffic…) from impacting subscriber’s QoE, operators need new network and service environments where policy systems are not limited to well-known scenarios but be capable of understanding and managing unforeseen circumstances.
The key feature for achieving that is coupling the Control Plane functions with Complex Event Processing (CEP) analytics engine that is capable of:
- Real-time analysis of information from networks, systems and applications. It is important to observe that the ingestion of additional non-network related information – such as hypervisor logs, application resource usage, virtual machine resource consumption, etc. – makes decision process more coherent and robust (i.g. making decision based not only on the traffic carried but also on the whole environment status) and allows to implement automated closed-loop decision process
- Identifying the best action for each customer’s session, using advanced analytics model and machine learning systems. i.g. new video customers can be associated to video servers not only using “network distance” but even latency, video server memory utilization and prediction of future load on that server.
Providing back the best action to the PCRF (in this scenario, the PCRF acts a close loop automated system). Analytics will be very useful to extract insights from network and cloud environments, but their main value will be in providing automatic recommendation on how to manage issues. Integrating real time and batch analytics with machine learning and recommendation engine will transform analytics in the intelligence of the networks and of the cloud environments, enabling real-time automatic close loops.
Complex Event Processing
CEP is a new paradigm to derive intelligence from event data in real time. From one side, it provides methods to define how events have to be processed and analyzed. From another side, it provides an analytic engine to aggregate, correlate and analyze events and patterns of events, producing new high-level event that can trigger an action.
Summary
The combination of Analytics and Virtualized Control and Data planes allows CSPs to realize networks and services environments capable of re-configuring themselves in order to manage high level of customer’s perceived experience within today`s economic constraints.
Coupling CEP capabilities with Policy’s will allow Operators to deploy self-adaptive networks and cloud environments capable of automatic close loop decision process and of subscriber’s tailored services.