IBM Cloud Satellite: A Wider Playing Field for the Cloud


IBM recently announced a partnership with Bharti Airtel in India to provide hybrid cloud services for edge computing. We take a look and interview IBM’s VP & CTO of Networking & Edge Computing.

A picture of a building with the IBM logo.
Image: nmann77/Adobe Stock

IBM and communications services provider Bharti Airtel have joined forces to provide Airtel’s edge computing platform to India. It will include 120 data centers across 20 cities, and demonstrates interest in edge computing deployed as a hybrid cloud environment.

This also anticipates the rollout of 5G in India next year, which is predicted to have an economic impact of $1 trillion by 2035, according to Telecom Regulatory Authority of India Chairman PD Vaghela.

Likewise, edge adoption and adaptation are growing. IBM VP & CTO of networking & edge computing Rob High points out that today’s crop of 15 billion edge market devices is only expected to grow.

Growing adoption of the hybrid cloud

Hybrid and multicloud enables organizations to take advantage of cloud-native development practices and abilities without being beholden to one or another of the large cloud providers.

SEE: Hiring Kit: Cloud Engineer (TechRepublic Premium)

“They need to be multicloud and be able to adapt to run their business at the speed of business,” High said in an interview with TechRepublic. “They want to be able to decide where to place their workload. They want to be able to decide whether to use a public cloud or their private data centers.”

Moving even closer to the edge, the problem becomes that each edge point is very different. It may have different architectures, different storage and network capacities, or be configured differently. Some edge points may be on the move, like smart vehicles. Multicloud addresses the diversity and dynamism of edge devices, High said.

Another element here is multi-access edge computing, which plays in the same arena as the cloud. Telecommunications companies often sell servers for clients with network capabilities on either a dedicated or private MEC. For Airtel and other telcos, that is part of their edge story.

“They’re trying to make compute capacity available to their enterprise clients, so instead of having to have a server in my store, I could simply put it into that central office, regional data center or hub location and offload the need to have the hardware in the store itself,” said High.

In this case, IBM is helping set up Airtel’s MEC infrastructure in a hybrid environment based on IBM Cloud Satellite and Red Hat OpenShift for their clients who want edge computing to reduce latency or transmission costs. It reduces the need for a lot of processing power in a factory or retail store.

Specifically, IBM contributed Edge Application Manager, a commercial distribution of the LF Edge Open Horizon Project — an open source project for which IBM is a primary donor. Airtel will pass this capacity on to Maruti Suzuki India, India’s largest passenger car manufacturer, which plans to use it to increase accuracy and efficiency for quality inspections on the factory floor.

Declarative systems

Where this differs from a single-cloud system provider such as AWS is that the clients can create different policies for different edge nodes without being beholden to a top-down system. Nodes can be set up differently depending on hardware such as whether it has a GPU or the number of cores.

The agent at the node will negotiate with an agreement bot at the central management hub location to establish what software to place at that node, High said. The agent and the agreement bot will continuously monitor those agreements and create a new negotiation if anything changes that invalidates the original negotiation.

“In this way, the system remains up to date, even in the presence of high rates of change,” High noted.

Another reason to give each node its own agency is for security. It doesn’t expose any open points to the outside world, High said. The agent automatically verifies the signatures attached to any workload to prevent tampering. Plus, the edge node can run even if the local network goes offline.

What’s next for edge computing?

“The fact that edge computing is emerging on the marketplace very rapidly is an issue for those organizations who are not looking at exploiting edge computing to their own advantage,” High predicts. The goal is to reduce operational cost by reducing the amount of organizations have to deal with at end points. It also solves the problem of how to deal with the growing complexity of non-IT staff being the front line of managing IT devices.

For Maruti Suzuki, that means leveraging IBM Maximo Visual Inspection for AI-enabled quality inspection on Airtel’s 5G offerings.

“We have the largest network of edge data centers available in India under the Nxtra brand and we will leverage our work with IBM to help Indian businesses address their critical business needs with greater efficiency, making it significantly easier for companies to process workloads where their data resides,” said Ganesh Lakshminarayanan, CEO Enterprise at Airtel Business, in a press release.

High notes that in general, organizations need to remember not to treat management or managing the edge as an afterthought. It isn’t possible or practical to have a person monitoring every edge device at all times. No matter whether you start with one device or hundreds, it’s key to get into practice managing them properly.