The shift to analytics-driven programmable networks
Communications service providers (CSPs) wanting to transform to new operating models must innovate across many areas of their business. They need to build agility into their operating model in order to make optimal use of their assets to provide digital services supported by an agile, virtualized next generation network.
CSPs are beginning this transformation to provide new digital services within their geographies - moving up the value chain and supporting these new services with new systems and changing their operating models in an iterative process.
For CSPs, their customers and networks are among their greatest assets. CSPs now need to compete on the basis of customer satisfaction, and the focus has moved to specific areas with measurable business impact like personalized offers.
Real-time network and operational data will play a pivotal role in enabling network automation and business transformation. CSPs now use this data to provide a better customer experience, mostly for network facing manual and automated optimization using SDN, and increasingly for customer-facing applications like personalized offers and advertising. Capacity management tools that use this data to provide automated just-in-time capacity improvements are also available.
Real-time data processing
CSPs have made significant investments in big data technology and systems. Real-time streaming analytics technologies such as Apache Kafka and Apache Storm enable CSPs to store, analyze and visualize real-time network data.
Real-time data processing engines that run in-memory with streaming and batch data are becoming available. Data architectures such as Lambda bring together batch-based approaches like Hadoop and streaming data to combine two different technologies’ high-speed data to be processed using low-cost data infrastructure.
CSPs need stable hardware and software platforms that are flexible, scalable and carrier-grade. They need an architecture that will enable them to dynamically to program and configure their network to suit their customers’ needs. Network innovation through cloud, NFV and SDN technologies will support and accelerate this service innovation by providing dynamic resources and the capacity to support highly scalable digital services.
However, a key requirement that will underpin these programmable networks will be the availability of the real-time data analytics that will provide the complete visibility of the network to the orchestrator through metadata. This is already recognized in Tier-1 CSPs’ proposed architectures such as AT&T’s ECOMP framework for managing its Integrated Cloud nodes, and NTT’s own orchestrator and architecture.
There are other examples of closed-loop, policy and analytics-driven programmable networks. SK Telecom is planning a business transformation strategy around big data analysis and network automation by coupling network monitoring coverage with SDN control. It expects to build new revenue sources like advertising and improve network utilization by increasing automation, eliminating redundancy, optimizing customer experience and streamlining the process.
The Telstra PEN platform is based on a centralized SDN controller that has end-to-end visibility of the network, and controls flow deployment including service chaining for NFV.
Orchestration will play the active role of scheduling resources for all applications subscribing to the network as a service and accessing the service dynamically through APIs. When the capability for end-to-end network orchestration is developed, this will enable network slicing - a key element of the 5G networks of the future.
Shanthi Ravindran is a senior analyst with Analysys Mason
This article was first published in Telecom Asia Big Data Insights April 2016 edition
Machine learning is another key technology to enable development of the sector