In recent years, enterprise workloads have rushed to the public cloud to take advantage of the cost-cutting, security, and data modernization benefits that a cloud-based platform provides. But Accenture is now helping many customers avoid running applications, and sending data, outside of their retail shops, groceries, manufacturing plants, and agricultural fields.
It’s not that those workloads are moving back from the cloud; it’s that the cloud is moving to those geographically dispersed facilities.
The advent of edge computing infrastructure—devices that run compute functions at the ‘edge’ of the internal networks within customer facilities—along with enabling technologies such as the Internet of Things (IoT), 5G, and application container software, have finally made it possible to effectively deliver cloud-native services directly besides the sources generating data.
Those advances usher in the notion of a distributed cloud, and the imperative of striking a balance between public cloud and edge infrastructure that optimizes application functionality and IT spend, according to Mrinalini Iyengar, global go-to-market and sales leader for Accenture’s edge computing practice. She shared her thoughts in a session at the Accenture Oracle Leadership Council, held in Austin, Texas, in June 2022.
“If you have data and databases, edge computing increasingly needs to be a component of your enterprise data strategy,” Iyengar told attendees.
Accenture is helping clients harness these emerging technologies and successfully bring them into play within their organizations, said Brian Daugherty, an Accenture senior manager, at the same session.
“Because we are using cloud capabilities all the way from cloud to edge, we’re really calling this the cloud continuum,” Daugherty said. While there’s no widespread consensus for where the edge starts in this ‘continuum’, he added, Accenture mostly defines the term as infrastructure outside of corporate data centers or public clouds.
There are unique advantages to running data-intensive applications, such as process control or computer vision, close to their data sources. But as the technology becomes more sophisticated, enterprises face new questions about what workloads should, or should not, be pushed to the edge of their networks.
“It’s not like we’re trying to leave the cloud and go back on-premises,” Daugherty said.
Accenture has been assessing customer use cases and evaluating which applications can effectively be delivered in edge environments. For example, those that must run in real-time, or close to it, are good candidates. “In some cases you need to make decisions and act upon data faster than it takes to get to the cloud and back again,” Daugherty says.
Bandwidth and storage considerations also come into play—gigabytes or terabytes of data streaming from arrays of sensors can run up data egress and storage fees. To minimize those expenses, Accenture implements solutions involving data transformation, contextualization, and filtering to reduce the amount of data sent to the public cloud.
Another motivator for edge deployments are government data residency regulations, or just concerns about sensitive data leaving the customers’ premises. Daugherty also mentioned mission-critical processes—such as industrial automation—that must remain running even if network connectivity is lost or degraded.
Accenture Managing Director Rita Gupta has been working with retailers on edge transformations, helping grocers and apparel shops get the most out of their investments. For example, many retailers have introduced self-checkout capabilities that speed the customer experience by reducing wait times. The next step is often to add computer vision for fraud detection and loss prevention, or stock monitoring and shelf analytics systems. Many retailers initially tried running those applications in the cloud, only to be hit by high data transit fees.
Edge infrastructure provides a more cost-effective option, helping front-line workers cope with labor shortages and evolving consumer demands following the global pandemic, Gupta says.
While these use cases are instructive, every company needs to carefully assess the unique capabilities and constraints of its IT environment before procuring edge systems. According to Gupta, customers should strive for pilot investments that are scalable and provide instructive feedback to avoid over-investment. “The key is to find that right balance between future-proofing your use cases and building an architecture that allows you to scale without creating these silos of infrastructure,” Gupta says.
It’s also important to remember that workloads travel a two-way street across the cloud continuum. At another AOLC session, Mike Sicilia, executive vice president of Oracle Industries, considered how these emerging technologies create new incentives to push some on-premises IoT applications out to the public cloud.
With 5G networks boosting wireless bandwidth, there can be value in “dumbing down on-premises devices, and moving much of their compute into the cloud,” Sicilia said. That approach makes it possible to replace expensive machines with less sophisticated sensors, which are also cheaper to operate and bring the added benefit of high-level consolidated control.
“You can change at any given time and automatically it’s reverberated through anything that’s connected,” Sicilia said.
In short, Sicilia said, “there’s a balance on how much compute belongs close to the edge and how much belongs to the cloud.”
Sicilia told the Oracle and Accenture customers who gathered for the AOLC conference that they’ll see this balance at play in new AI, robotics, and shop-floor automation products Oracle plans to unveil this year.
Joe Tsidulko is a senior communications director at Oracle, who previously covered enterprise tech for CRN, and long before that, crime and criminal justice.