By Stephane Honore*, Chief Technologist,
Openwave Mobility
In the previous
post, we described the importance of driving clear ownership and accountability of user data policy and management within the operator organization to truly achieve SDM new monetization opportunities. In this second part, we will see how the trend of centralizing user data from an organizational and political perspective is being reflected in the realms of technology.
Challenged of federated models
Many operators have taken the “quick fix” approach of data federation to provide a virtual logical view of user data and to hide the fact that data is actually stored in a multitude of separate back-ends. The main driver for this implementation is to provide the least disruptive path for achieving logical user data consolidation across the network.
However when such solutions are considered permanent, they jeopardize the long-term ability of the mobile broadband operator to truly transform their user data into monetization opportunities. Hybrid or federated consolidation models have the following fundamental short-comings:
- They do not solve the root cause of the data everywhere problem: operators have simply added plumbing software layer on top of their silo’ed services and back-ends. Under the hood no consolidation has actually taken place, neither in the infrastructure nor at the organizational level.
- Dealing with multiple data backends who have different data models requires complex data processing capabilities to be built in the federation middle-ware; in addition data provisioning and retrieval flows have to take into account failure conditions through challenging roll-back procedures prone to leaving dirty data in the midst
- Provisioning complexity and lack of true consolidation eventually leads to increased latency as a single request might end up being forked into multiple sub-requests to distinct back-ends before all the replies get compiled into a single response
- It is not possible to guarantee the overall performance of the solution since the weakest element of a federated infrastructure becomes the weakness of the entire system
3GPP originally advocated the federated SDM models through the GUP server standardization that provided a middle-ware layer responsible for the implementation of data retrieval logic (to determine which back-end a request should be routed to) and access-control definition (to determine who can get which data). Luckily SDM solutions went through a maturation process over 5 years to arrive at the conclusion: true data centralization cannot be realized logically and requires abstraction between the applications and the data store.
Towards User Data Convergence
After having tackled the user “data everywhere challenge” for a number of years, 3GPP eventually settled on a simple architecture known as User Data Convergence (UDC) with the introduction of the User Data Repository component (UDR) and the definition of the Ud reference point towards external application front-ends. The work started in 2009 with the definition of the service requirements (3GPP TR22.985) supported by major Tier1 carriers and reached completion in Rel10 with the UDC technical specification (3GPP TS-29.335) standarizing the UDC architecture. Since then only minor improvements have been added to the now solid and stable UDC foundation, including:
- 3GPP TS-32.182 started the definition of objects and relationship (end-user, subscription, servce profile, identifier) towards a Common Baseline Information Model (CBIM)
- 3GPP TR-32.901 studied the UDC information model handling and provisioning to develop understanding, guidelines, and preliminary requirements for the BSS provisioning capabilities and information model handling when new applications and related services are implemented in the operator’s UDR
As Diameter is becoming a popular transport protocol in mobile broadband architectures, operators also have the option to use a Diameter based interface against the UDR with the 3GPP Sh interface (3GPP 29.329) which initially was used to exchange user profile, location information, and charging function addresses in the IMS realm (between SIP AS and HSS).
Compared to the federated models, the 3GPP User Data Convergence model provides an efficient, mature and standardized way for achieving user convergence without any middle-ware layer as long as front-ends and the UDR back-end talk the same Ud/Sh protocol, something any mobile broadband operator willing to adhere to the UDC deployment model should enforce on vendors. Given the duality between LDAP and Diameter, it is important to takes a best of breed approach with a UDR capable of supporting both stacks on the same software component to avoid the need for deploying new software or even new hardware when the operator wants to make use of the Sh interface in the future.
Think “push” rather than “pull”
It is often overlooked that the 3GPP UD interface specifications give operators more than a simple data storage and retrieval mechanism. While the main purpose of the interface is to allow any front-end (service or application platform) to query, create or update the user data information over a standard LDAPv3 interface, 3GPP also defined a new synchronization and replication mechanism by permitting front-ends to subscribe to any user data changes happening in the UDR and being notified over a standard HTTP/SOAP interface.
This is an important aspect of the UDC architecture as it goes beyond the traditional pull mechanism for getting updated user data information. This notification mechanism based on a subscription/push model is a game changer as it provides a simple, yet efficient, way to keep local caches/stores synchronized with the centralized UDR master. It provides the possibility of easily implementing new use cases that require a real time view of the subscriber data. As immediacy and instant gratification becomes the standard of Internet services (1-Click real-time experience), mobile operators need to look at new ways to provide real-time self-care capabilities where user self-care interactions immediately take effect on their traffic management or charging infrastructure. Examples of such services include roaming plan sign-up or application-based charging top-up scenarios where service activation immediacy is required.
With 3GPP push-based architectures operators can now consider launching new services that have strong real-time requirements between user interactions and network infrastructure.
Conclusion
Making user data convergence part of the network modernization agenda and executing organizational adjustments to clarify ownership and responsibility of data across the organization are key commitments operators need to make to fully benefit from SDM solutions. Failing to do this they won’t be able to set the growth for the future without solid subscriber data management foundations.
___________
*Stephane Honore is Chief Technologist for Openwave Mobility with over 15 years’ experience in the telecommunication industry, responsible for contributing to the technology and product strategy of Openwave Mobility’s mobile internet products. He is also responsible for the SDM product management. Prior to his current role, Stephane held a wide range of leadership roles within the Sales, Engineering and Product Management organizations