Your browser is not supported. Please update it.

27 November 2019

Vodafone dumps inhouse analytics system in favor of Google Cloud

Google has been rather overshadowed in the telecoms cloud world by AWS and Microsoft Azure, but it is starting to gain ground with some significant operator partnerships, often driven by its AI (artificial intelligence) expertise. Hard on the heels of an alliance with Telecom Italia (see Wireless Watch November 13 2019), it has announced a similar arrangement with Vodafone.

Many operators are turning their back on inhouse cloud and AI developments, in favor of taking advantage of the webscalers’ heavier investments and more scalable solutions in these areas. Vodafone has been using its own on-premises system, based on the Hadoop open source technology, to analyze business data across 11 countries, but that system requires eight clusters of servers (600 in total), and is still creaking under the strain of the big data requirements.

So Vodafone is now going to use the Google Cloud Platform for AI-enhanced analytics, in a step which may lead to further applications, and even network functions, moving to the platform in the future.

The two companies have co-developed a customized system called Neuron which runs on the Google cloud, and all Vodafone’s global data will be migrated from its current home, spread across numerous ‘data lakes’, to a unified ‘data ocean’, as the operator described it.

The company’s group CTO, Johan Wibergh, told Bloomberg that Neuron would be more advanced and more scalable than the inhouse Hadoop platform, but acknowledged that Vodafone lacked the internal skills to build Neuron alone – and even if it managed to hire them, that approach would have been far more expensive.

Simon Harris, Vodafone’s head of big data delivery, wrote about the new partnership in a blog post. “Neuron serves as the foundation for Vodafone’s data ocean and the brains of our business as we transform ourselves into a digital tech company,” he wrote. “Not only will we be able to gain real time analytics capabilities across Vodafone products and services, it will also allow us to arrive at insights faster, which can then be used to offer more personalized product offerings to customers and to raise the bar on service.”

Tasks which could be enhanced by Neuron, and which would fundamentally affect 5G and converged network performance and cost-effectiveness, are key. For instance, the engine could be used to predict traffic spikes and allocate capacity to prevent a poor customer experience. It could also be used to personalize services for each subscriber and understand their real world network experience in real time.

The use of AI is rising among advanced operators, as a way to achieve their ultimate goal of a zero-touch network while ensuring that quality of experience is maintained. Vodafone is targeting a net reduction in annual operating expenses of at least €1.2bn ($1.3bn) by the 2021 fiscal year, and full automation will be one element of that, as will the migration of more functions to the external cloud.

Wibergh denies that this Google alliance will open the door to selling off some or all of Vodafone’s data centers, even though several large operators have already gone down that road, including Verizon. These carriers have admitted that building their own cloud data centers, for internal use or to support customer services, is not their core expertise, and that they cannot achieve the cost efficiencies or skills base of the cloud specialists. But Wibergh says that, in Vodafone’s case, the data centers still have other active functions.

And some operators, such as Deutsche Telekom, still insist that there is a role in the value chain for the telco cloud, especially because of public concerns about data privacy and security on the major webscale platforms. And others are building their own AI tools too, notably Telekom Austria, whose systems maybe adopted by its controlling stakeholder, Latin American giant América Móvil.

For the operators which are moving to the public cloud, there will probably be a move to have multiple partners, rather than moving the network lock-in, about which they have complained for so long, down to the cloud infrastructure. AT&T has strategic relationships, targeting different aspects of the business, with Azure, AWS and IBM, as well as its own cloud platforms, for instance. In a distributed group like Vodafone, there may be local cloud partner choices for each operating company’s data, as well as a group-wide platform and partner.

To improve its appeal to telcos, Google Cloud has launched a Network Intelligence Center to provide visibility and control for cloud-based virtual networks.

“Cloud visibility has been a black box so far,” Shailesh Shukla, Google Cloud VP of products, told Light Reading. But there is a requirement to manage huge numbers of microservices and containers, and their connections, which will be turning up and down and reconfiguring themselves rapidly.

Network Intelligence Center aims to provide better, simpler visibility into how workloads and microservices communicate with each other globally in multicloud environments. It provides monitoring, verification and optimization across the cloud and on-premises data centers, Shukla wrote in a blog post.

There are four modules – Connectivity Tests and Network Topology, both in beta; and Performance Dashboard and Firewall Metrics & Insights, both in alpha, with other units to follow.