Policy servers must be scaled up

The policy server, also known as the policy and charging rules function (PCRF), is essential for service providers to adopt a broader set of strategic policy use cases. These new use cases focus on personalization and adding value to cloud, over-the-top, and machine-to-machine (M2M) services, as well as a more sophisticated approach to managing network resources.
The significance of the policy server will continue to grow as service providers migrate to LTE and offer voice and video over LTE. These services will increase Diameter signaling traffic and require application prioritization to ensure quality of service (QoS). The proliferation of M2M devices and applications and new cloud services will further elevate policy’s role as operators prioritize services to meet service level agreements and QoS requirements.
The capabilities and flexibility of any policy server depend on the number of distinct triggers, conditions and actions that it can process. These, in turn, rely on the number of devices with which it can interoperate. The policy server’s capabilities, therefore, grow exponentially by increasing the number of devices it supports and by expanding the scope of functionality on the interfaces between it and other enforcement points. Its role is to create and apply policy rules based on a variety of parameters – subscriber profiles, application usage, device type, and network conditions – that are distributed to policy endpoints on the network, which control the flow of traffic.
Policy integration with complex charging and billing systems is a particular pain point for operators. Heavy Reading notes that the single-biggest barrier that operators face in deploying a policy architecture is the difficulty in integrating policy and charging. The analyst firm also found that operators have run into major challenges implementing online charging systems due to the complexity and high costs associated with customization, professional services, and integration with gateways and policy servers. Less than a tenth of the operators intend to deploy pre-integrated policy and charging solutions in the medium term to solve integration issues.
The 3GPP architecture specifies several interfaces – Gx, Gy or Gz – to online and offline charging systems (OCS/OFCS) through the policy and charging enforcement function (PCEF). However, the 3GPP doesn’t define how this integration is implemented in practice.
In Release 11, the organization defines a new interface, called the Sy interface, to create a direct link between the PCRF and charging systems. The Sy interface eliminates the need to do usage tracking over both the Gx and Gy interfaces simultaneously, improving messaging efficiency and eliminating the possibility of accounting synchronization problems between the PCRF and OCS. Most importantly, it eliminates the need for complex and costly pre-integration between the policy server and OCS. Because of the critical importance of this link, support for the Sy interface is a key criterion for service providers selecting a policy server.
Service providers will need to create more policies, using a wider range of triggers, than they do today. They’ll need to chain or “nest” policies together like customer tier and application or application and device to create the complex policies that deliver services.
As the number of policy use cases increases, outsourcing their creation to third-party vendors will become cost-prohibitive. It’s also an issue of time. The timeline for third-party development and professional services can be a year or more to create and test a single policy. It’s essential for service providers to have the ability to respond quickly to changing network conditions and subscriber behavior with a flexible and agile policy creation environment. 
Key features of easy-to-use policy creation tools are a user-friendly graphical user interface that enables service provider personnel – even non-specialists – to intuitively engage the system, and system interoperability with support for standardized and non-standardized interfaces for quick and easy integration with other policy endpoints. They also include a diverse menu of policy triggers to create advanced rules, and integrated policy analytic tools that enable operators to understand traffic patterns and user and service behavior.
Policy and the new Diameter network
Industry projections point to dramatic increases in both mobile data traffic volume and data-enabled devices and applications. They will also result in a dramatic increase in Diameter signaling traffic, which is expected to exceed data traffic, among policy servers, charging systems, subscriber databases, and gateways. Diameter serves as the interface between numerous 3G and 4G network resources involved in policy negotiation, including GGSNs, HRPD serving gateways, mobility management entities, home subscriber servers, OCSs, OFCSs, PCRFs, and application servers.
Two of the most critical factors to weigh when evaluating the scalability of a policy server are expertise in Diameter routing as well as networking and carrier-grade functionality, including high TPS, processing power, capacity, overload controls, monitoring, and protection mechanisms. 
Many charging and major network equipment vendors employ proprietary interfaces or vendor-specific implementations of the Diameter protocol. This approach creates interoperability problems and hampers service provider efforts to scale their policy and charging systems to support network growth.
In the absence of a new Diameter signaling architecture, each endpoint must have a direct Diameter association with every node to which it is connected, creating a mesh-like network architecture. As data and signaling traffic levels swell, the lack of a capable Diameter signaling architecture poses a number of challenges.
One is scalability. Each endpoint must maintain a separate association with all of its Diameter peers as well as the status of each. This requires every element to handle all session-related tasks such as routing, traffic management, redundancy, and service implementation.
Another is interoperability testing (IOT). Protocol interworking becomes unmanageable as the number of multi-vendor devices increases. With no separate signaling or session framework, costly and time-consuming IOTs must be performed at every existing node when a new node or software load is placed in service.
Then there is PCRF binding. When multiple PCRFs are required, there must be a way to ensure that all messages associated with a user’s IP connectivity access network sessions are processed by the same PCRF. 
And then there is protocol mediation. There’s no centralized point in the network from which to mediate the different Diameter protocol variants.
Yankee Group reported that the impending explosion of Diameter traffic that will follow LTE and IMS deployments will require new policy signaling architectures. Anticipation of this growth will ensure that next-generation networks are built for scale from day one and will not have to endure expensive retrofitting or risk potential signaling-layer failure and sub-optimal user experience.
The selection of a policy server is one of the most critical decisions service providers face in their quest to monetize and manage the mobile data traffic. The ability to deliver complex policy use cases, while managing swelling traffic and signaling loads, requires an advanced policy server that is highly interoperable, can rapidly create new services, and delivers high performance and scalability as part of a new Diameter architecture. 
Doug Suriano is CTO and VP of engineering at Tekelec