Feature

Living on the edge: The enterprises working with distributed cloud

Distributed public cloud services are seen as driving future advances at the network edge but few companies are as yet enjoying full-fledged benefits

This article can also be found in the Premium Editorial Download: Computer Weekly: The jobseekers turning to cyber crime in the pandemic
Fleur Doidge
By
Published: 20 May 2021

By 2025, most cloud service platforms will provide at least some distributed cloud services that execute at the point of need, according to Gartner, for a cost-effective single pane of control while staying the right side of data sovereignty and residency requirements even at the edge.

Dennis Lauwers, distinguished engineer for hybrid cloud at IBM Europe, Middle East and Africa (EMEA), gives the example of deploying artificial intelligence (AI) capabilities at a branch or edge location – without managing the whole AI engine.

“I can have the same cloud experience as public cloud in the location I need,” he says. “The service is activated for me and consumption-based, with ease of use, managed and monitored. I can do a pay-as-you-go or pay-as-you-use model because of that.”

Gartner positions distributed cloud partly as an answer to hybrid complexities and public cloud compliance concerns, with Gartner senior director analyst John McArthur recommending distributed cloud as a solution for edge. However, beyond the likes of Google Cloud and Anthos on-premise, adoption presently looks embryonic.

Lauwers says IBM partner Lumen is bringing cloud to edge locations for industrial internet of things (IIoT) applications, consuming capabilities in a distributed way to serve Lumen’s own customers.

“Their solution is now built on public cloud services but they can deploy practically to any environment today they really want,” he says.

“We also have an ISV [independent software vendor] working with retailers bringing software to the retail locations, not their datacentres, to do very nice visual inspection things and so on, ensuring everything is running optimally into a store, with a runtime environment to automate application delivery into those locations.” It’s an approach he likens to “software as a service on-premise”.

Organisations are certainly provisioning resources and linking into a manageable control plane, and IBM’s offering could even run enterprise SAP at the edge “if you really needed to”, Lauwers says.

IBM Cloud Satellite can combine IBM and Microsoft Azure services, for example, for a lower latency solution that executes on location and enables data governance and sovereignty within a “catalogue” cloud service, he notes.

Gordon Davey, a global cloud services lead at SoftwareONE, agrees that distributed cloud with improved transparency should attract many organisations, as they move away from the idea of cloud being simply “an amorphous blob” for storing (or losing) your data.

SoftwareONE customer Omnico, an Azure-based transaction and engagement platform provider to global retail, food and leisure brands, managed to halve cloud costs while preserving scalability and support in an automated environment with multiple nodes.

The Omnico platform has improved availability and security, including better ISO27001, General Data Protection Regulation (GDPR) and ITIL compliance, Davey says.

“Acquiring that flexibility, combined with new innovations like confidential computing, can drive higher levels of encryption and higher data security within specific regions,” he says.

Large-scale projects especially can gain from deploying Azure or similar for customer environments in search of granularity and feature parity. Organisations can also trial distributed cloud in small bites to “fail fast or scale”, Davey adds. 

Stephen Gilderdale, senior director and head of presales architects at Dell UK, confirms that use cases are emerging with Dell partners both for “near edge”, with firms such as Equinix for colocation or Orange, and “far edge” IoT sensor-based applications for “wherever the work is taking place”.

Distributed edge cloud potentially answers analytics questions faster, helpful for matters such as moving goods or getting on-site permissions.

Dell has been talking to large shipping and logistics firms about ensuring the right freight gets onto the plane or ship going to the right location to deliver items economically and fast, as well as distributed cloud use cases with football stadia and construction companies, involving cameras, computers, tools and equipment – including radio-frequency identification (RFID) tags and the like for site visitors.

“We were to do one particular football game managed by partners, having cameras check people’s temperatures and the flow of people. The game got cancelled [due to lockdown] but trials will continue,” reveals Gilderdale.

“We’ve been discussing with construction consortia how to build edge at larger building sites, such as new shopping centres – architects and builders on site follow the latest updates to documents and specifications as they work. That is a real-world opportunity.”

Organisations might hang back initially to see what rivals do that generates competitive advantage, Gilderdale adds.

Other adoption blockers

Srinivasan CR, chief digital officer at Tata Communications, agrees that many firms have begun considering distributed cloud, with it typically being about dynamically updating data collected at an edge location where latency and flexibility matters.

“Edge computing is becoming a factor in every solution that customers are thinking about – whether the compute can be close enough to where the functionality needs to be delivered,” he says.

Analysing frames of video on a factory floor, for example, enables proactive notification of safety breaches and adherence to protocols across a site. In retail, edge analytics can monitor what interests people in-store. Right now, however, even large organisations tend to focus more on private-public hybrids rather than distributed public cloud services.

“The uniformity of a distributed cloud – with public cloud locally available in various locations – would help, but we are not seeing too much of that at this time, either as a consideration or as presence,” Srinivasan explains.

Uniformity of management and billing is desired, yet the hyperscalers may need to simplify their offerings for on-premise, which might make it difficult to run the edge applications businesses want. Meanwhile, many enterprise applications remain un-containerised, Srinivasan points out.

“But I think the stratification of cloud is going to happen: the hyperscalers, the telecoms cloud, the edge cloud, the on-prem cloud. But we’ll have to wait and watch,” he says.

Dan Davies, CTO of cloud communications provider Maintel, agrees that the ball remains mostly in the cloud giants’ courts to produce suitable offerings.

“We still see a huge demand for private cloud, because we can deploy that in hybrid mode with some elements on-premise, some in the private cloud, integrate them and have that kind of managed environment,” he says.

That said, distributed mobile and cloud deployments will increasingly aim to move compute towards 5G networks to support advanced edge applications, with economies of scale still needed to lower the price point.

“It’s easier in a smaller number of really large locations, but then we hit those challenges around data sovereignty and latency and so on. We saw a trend of organisations starting to pull payloads back on-premise – and, of course, the public cloud providers really don’t want that business taken away,” Davies adds.

David Friend, co-founder and CEO of cloud storage specialist Wasabi, says the Googles, Amazons and Microsofts of this world are so dominant as to reduce incentive for other providers – yet customers drowning in data are seeking cost-effective solutions.

“You can run your application wherever you want, whether 10 years of hospital X-rays and MRI, or data from a satellite or a telescope in Chile, or genomic data and energy exploration data,” says Friend.

James Mernin, director of engineering for cloud services at Red Hat, says hybrid cloud still suits many customers even when it means multiple architectures and divergent applications. Its OpenShift service built on Kubernetes runs natively on Amazon Web Services (AWS).

“Our customers want to keep some of their infrastructure, but also want to have Red Hat run some cloud elements for them,” Mernin says.

Emilio Valdes, EMEA senior vice-president at cloud data manager Informatica, agrees that more companies seek the flexibility and other advantages of cloud with more transparency and control, especially since the onset of the Covid-19 coronavirus pandemic.

“Many are facing challenges to connect multicloud environments,” says Valdes. “Sometimes business users struggle to find the relevant data. They don’t know where it is or if it’s in one cloud, private or public.”

Read more about distributed and edge compute environments

Read more on Datacentre capacity planning