Tony Jackson explores whether innovative approaches to policy control and charging, combined with network functions virtualisation, can give digital service providers the agility they need to meet customer expectations
The author, Tony Jackson, is senior product manager, Policy Control & Charging at CSG InternationalNetwork Functions Virtualisation (NFV) and Software Defined Networks (SDN) are expected to drive operator agility and cost savings. But technologists and service providers alike need to carefully consider the implications of virtualisation. One candidate for a deeper look is policy control and charging systems (PCC). As communication service providers (CSPs) transform into digital service providers (DSPs), and explore the possibilities engendered by NFV, what’s next for policy control and charging?
First, it’s important to recognise policy control and charging has evolved dramatically from a defensive function utilised to control scarce network resources, to a more dynamic application aimed at monetising emergent data services. The days of crude bandwidth throttling have been overtaken by a more positive approach focused on retaining – and even delighting – customers, who can now enjoy on-demand bandwidth boosts for video, social networking usage included in base subscriber packages, bill shock prevention through roaming notifications and many more policy-driven applications designed to improve their experience and increase their loyalty to the CSP.
We expect this trend to continue as LTE, mobile broadband and the internet of things become more widespread, creating opportunities for innovative service providers to develop new revenue streams and business models, individualised for their customers. With VoLTE finally coming of age, policy control and charging will be extended to help monetise HD voice, as well as promote enterprise plans and control QoS. Tightly integrated policy control and charging will be essential tools for DSPs in the age of the customer.
Innovation and agility
When did you last hear your marketing team say: “This month, we launched 18 new packages – 14 for smartphones and four for Wi-Fi – across post-paid, pre-paid and hybrid payment types. Next month, we’ve planned more than a dozen more, because it takes us only 15-20 minutes on average to configure, test and launch.”
For most CSPs this kind of agility and responsiveness is fantasy – but some are demonstrating this is achievable.
Agility, perhaps 2014’s buzzword of the year, is the measure of a company’s quickness and its ability to respond flexibly to a volatile market. Policy control and real-time charging is typically central to this agility. Tightly integrated policy control and charging allows CSPs to quickly configure new service packages, enhanced with intelligent personalisation, and enables the two-sided business models that are gaining momentum and are expected to typify the digital service portfolio.
A provider’s policy control ecosystem is defined by the different use cases that it must support. Some require simple integration between policy control and charging elements while others require interaction with other service components and functions, such as subscriber and traffic analytics, video optimisation and network congestion tools in a more complex environment.
Virtualisation for real cost savings
Recent research by Telecoms.com Intelligence shows that nearly 25% of CSPs are trialling virtualised BSS operations and roughly 15% have begun to, or plan to, deploy virtualised BSS this year. Virtualisation and cloud-like approaches to deployment offer CSPs significant savings across the board. Operational and capital savings include shared computing resources, more efficient energy consumption, reduced carbon emissions, a widely available skilled resource base and COTS hardware. When this is added to the likelihood of greater service agility and easier innovation, the argument appears to be overwhelmingly in favour of NFV and SDN.
But simply being able to deploy in a virtualised infrastructure is only the first step. To truly realise the benefits promoted by NFV, CSPs will need to use the automation capabilities that are fundamental to any virtualisation and cloud strategy. PCC vendors will need to support this by tightly integrating their applications with the orchestration layer of the virtualised environment. Vendors with proprietary or heavily customised, closed application stacks will struggle to adapt to this new paradigm.
Virtualisation: A brave new world
Policy control and charging has a direct impact on the subscriber’s end user experience, but virtualisation can impact the underlying infrastructure’s ability to meet real-time requirements. If multiple virtualised functions share physical hardware with policy control and charging, for example, it can introduce unacceptable latency – a critical performance factor which will directly impact the subscriber’s experience. If the network element has to communicate with software that has moved to the public cloud, this can result in the replacement of LAN-like performance with slower WAN-like performance. In addition, the infrastructure available in some emerging markets may not provide sufficiently reliable or numerous connections to the public cloud to make such an approach even viable.
Security can also be a concern. Many PCC systems have access to subscriber-sensitive data that is subject to close legislative control. In the EU, for example, privacy laws apply to all member states but each individual state can pass overriding laws. This will influence corporate policy concerning where data is held and accessed, and may in some cases prevent a full adoption of virtualisation and cloud-based processing.
From a performance perspective it is generally preferable that real-time processing is co-located with the data it needs to access. This can be difficult to ensure in fully virtualised environments, particularly if the volume of data is significant. Also, traditional RDBMS (relational database management systems) may not always lend themselves well to virtualisation – virtualised storage must be able to deliver the throughput that these databases require and traditional databases may not support an elastic scaling model consistent with a cloud-based approach. Finally, the software licensing model of the database vendor may prevent a cost-effective approach when deploying in virtualised environments, with some vendors trying to enforce specific hardware platforms with restrictive licensing practices.
A new class of in-memory database is emerging that is better aligned with the capabilities of virtualisation and cloud, but CSPs need to be aware this a complex area. Some databases sacrifice transactional safety or functionality in favour of distributed scale, an approach which may not be suitable for true charging systems where the customer balance is sacrosanct. CSPs will need to make choices about what level of transactional safety they require across their application stack and where they may be able to compromise. In addition, CSPs must also consider the end-to-end effect, as selectively virtualising applications can have consequences across the entire stack that will need to be factored into any decision making process. One size fits all will not apply.
Policy control and charging: What’s next?
Policy control and charging has matured rapidly from a stand-alone network control mechanism to a fully integrated revenue-generating tool, helping CSPs to not only manage and optimise network resources but to encourage service take-up and usage. At CSG, when we think about what’s next for policy control and charging, we believe the next evolution will see more agility, innovation and cost-savings as digital service providers undertake network virtualisation.
The case for switching to a virtualised environment will depend on many factors, including investment in and status of the current platform, customer expectations and security concerns. In the longer term, however, cost and efficiency benefits make it seem unavoidable that what can be virtualised will be virtualised.
