logo_header
  • Topics
  • Research & Analysis
  • Features & Opinion
  • Webinars & Podcasts
  • Videos
  • Dtw

Telcos need an adaptive architecture to target enterprises

Market disruption has created uncertainty in areas of demand, opportunity, geopolitical alliances and regulation. For CSPs, ubiquitous coverage and availability are fundamental. But now, the transformation to a platform model is a business imperative for CSPs to become more agile and able to react with the required speed to turn uncertainty into differentiation and revenue.

Iain Harfield
19 Aug 2021
Telcos need an adaptive architecture to target enterprises

Telcos need an adaptive architecture to target enterprises

I was in a technical meeting recently when the head of network architecture and strategy for a global mobile operator said, “Telcos don’t want to be telcos anymore.” While this sentiment is not new, it reflects the impact of the changing marketplace that has eroded communications service providers’ (CSPs’) traditional revenue streams and how over-the-top (OTT) providers have raised expectations for customer experience.

Market disruption has created uncertainty in areas of demand, opportunity, geopolitical alliances and regulation. For CSPs, ubiquitous coverage and availability are fundamental. But now, the transformation to a platform model is a business imperative for CSPs to become more agile and able to react with the required speed to turn uncertainty into differentiation and revenue.

For the CSP, revenue growth lies with the enterprises that will drive IoT-based opportunities. To take advantage of these opportunities will require an agile, component-based architecture deployed on hybrid (physical and virtual) infrastructure that will connect network, IT, and enterprise data sources and deliver value from them. At TIBCO, we refer to this as “digital adaptation of the network core”.

An adaptive architecture enables a flexible and agile approach to how standard functions are deployed and gives CSPs more choices in selecting vendors. Consider standard functionality such as 4G/5G network policy and charging. These functions typically require data and event integration to the network core plus subscriber/customer data to make policy and charging decisions. In traditional architectures, they would most likely be considered single architectural components and acquired from specialist vendors.

In the decoupled architecture, these functions can be separated to great effect. For example, policy can be separated into distinct IT and network components, with the former handling subscriber policy decisions and the latter making policy-specific network decisions. The two are decoupled but integrated. The benefit is that each IT stack or different type of enterprise consumer can have its own and completely different policy, charging or any other functionality without impacting the CSP network core.

Decoupled architecture increases flexibility

Source: TIBCO

New telco requirements


An adaptive architecture is more important now because demand is growing faster than ever before for delivery of network-as-a-service and telco-as-a-service to enable standardization and B2B2C opportunities. There are some very challenging functional and non-functional demands placed upon such an architecture. Following are minimum required capabilities that are needed in the composable applications across mobile network core and edge:

  • API-led integration and management: allows integration flows and their monetization to be defined and reused by multiple parties inside and outside of the CSP.

  • Geographic high availability, low-latency data and session state: allows low latency performance and consistency for business decisions at network speed.

  • Data pipelining, event routing and filtering: allows treatment and routing of device and IoT event data, for edge and enterprise customers.

  • Low-latency transactional event processing: ensures consistency and durability of device and IoT events to assure accurate charging and decisions

  • Analytic and machine learnt stream processing : allows machine learnt algorithms to execute in an event stream making predictive decisions very close to the data sources.

  • Model operations (Model Ops): provides governance and automation mechanisms to rapidly deploy and re-deploy models into an event stream reducing algorithmic risk.

  • Low code, no-code development: enables faster and scalable development teams for unpredictable business logic and emerging use cases.

  • Hybrid private and public cloud deployment; protect the CSP as with multiple environments. These environments will change and evolve creating their own legacies.


For the work I’ve been involved with, processing the high volume of network event streams clearly requires scale. However, scale is not enough.

The treatment of events in a composable environment to handle use cases such as call setup, profile updates and the execution of rules-based policies whilst data is inflight must be executed with very low latency and with very high availability demands. It has been crucial for the success of the network teams to enable agile business value and revenues that are API-led and event-driven. This is especially important for enterprise use cases.

Enterprise needs the edge


Edge computing is an essential capability for CSPs to serve enterprises. The word “edge” is used a lot these days, but it boils down to putting capability closer to the data sources to address latency and the localized handling of data volume that will be driven by 5G enterprise opportunities and emerging private networks. In effect, it means reducing decision latency and data volume forwarding into the network core and cloud.

Enterprise use cases will be harder to enable than the traditional subscriber-based business model because CSPs cannot predict what they will be, their scale, the service levels needed or how to charge. The flexibility to adapt to these unknowns drives the architectural requirements of the modern CSP and the components within it.