The Multiplier Effort of Edge Computing

Companies capture more data and compute capacity at the edge. At the same time, they are laying the groundwork for a distributed enterprise that can capitalize on a multiplier effect to maximize intended business outcomes.

The number of edge sites — factory floors, retail shops, hospitals, and countless other locations — is growing. This gives businesses more opportunity to gain insights and make better decisions across the distributed enterprise. Data follows the activities of customers, employees, patients, and processes. Pushing computing power to the distributed edge ensures that data can be analyzed in near real time —a model not possible with cloud computing.

With centralized cloud computing, due to bandwidth constraints, it takes too long to move large data sets and analyze the data. This introduces unwanted decision latency, which, in turn, destroys the business value of the data. Edge computing addresses this need for immediate processing by leaving the data where it is created by instead moving compute resources next to such data streams. This strategy enables real-time analysis of data as it is being captured and eliminates decision delays. Now the next level of operational efficiency can be realized with real-time decision-making and automation. At the edge: where activity takes place.

Industry experts are projecting that 50 billion devices will be connected worldwide this year, with the amount of data being generated at the edge slated to increase by over 500% between 2019 and 2025, amounting to a whopping 175 zettabytes worldwide. The tipping point comes in 2025, when, experts project, roughly half of all data will be generated and processed at the edge, soon overtaking the amount of data and applications addressed by centralized cloud and data center computing.

The deluge of edge data opens opportunities for all kinds of actionable insights, whether it’s to correct a factory floor glitch impacting product quality or serving up a product recommendation based on customers’ past buying behavior. On its own, such individual action can have genuine business impact. But when you multiply the possible effects across thousands of locations processing thousands of transactions, there is a huge opportunity to parlay insights into revenue growth, cost reduction, and even business risk mitigation.

“Compute and sensors are doing new things in real time that they couldn’t do before, which gives you new degrees of freedom in running businesses,” explains Denis Vilfort, director of Edge Marketing at HPE. “For every dollar increasing revenue or decreasing costs, you can multiple it by the number of times you’re taking that action at a factory or a retail store — you’re basically building a money-making machine … and improving operations.”

The multiplier effect at work

The rise of edge computing essentially transforms the conventional notion of a large, centralized data center into having more data centers that are much smaller and located everywhere, Vilfort says. “Today we can package compute power for the edge in less than 2% of the space the same firepower took up 25 years ago. So you don’t want to centralize computing — that’s mainframe thinking,” he explains. “You want to democratize compute power and give everyone access to small — but powerful — distributed compute clusters. Compute needs to be where the data is: at the edge.”

Each location leverages its own insights and can share them with others. These small insights can optimize operation of one location. Spread across all sites, these seemingly small gains can add up quickly when new learnings are replicated and repeated. The following examples showcase the power of the multiplier effect in action:

• Foxconn, a large global electronics manufacturer, moved from a cloud implementation to high-resolution cameras and artificial intelligence (AI) enabled at the edge for a quality assurance application. The shift reduced pass/fail time from 21 seconds down to one second; when this reduction is multiplied across a monthly production of thousands of servers, the company benefits from a 33% increase in unit capacity, representing millions more in revenue per month.

• A supermarket chain tapped in-store AI and real-time video analytics to reduce shrinkage at self-checkout stations. That same edge-based application, implemented across hundreds of stores, prevents millions of dollars of theft per year.

• Texmark, an oil refinery, was pouring more than $1 million a year into a manual inspection process, counting on workers to visually inspect 133 pumps and miles of pipeline on a regular basis. Having switched to an intelligent edge compute model, including the installation of networked sensors throughout the refinery, Texmark is now able to catch potential problems before anyone is endangered, not to mention benefit from doubled output while cutting maintenance costs in half.

• A big box retailer implemented an AI-based recommendation engine to help customers find what they need without having to rely on in-store experts. Automating that process increased revenue per store. Multiplied across its thousands of sites, the edge-enabled recommendation process has the potential to translate into revenue upside of more than $350 million for every 1% revenue increase per store.

The HPE GreenLake Advantage

The HPE GreenLake platform brings an optimized operating model, consistent and secure data governance practices, and a cloudlike platform experience to edge environments — creating a robust foundation upon which to execute the multiplier effect across sites. For many organizations, the preponderance of data needs to remain at the edge, for a variety of reasons, including data gravity issues or because there’s a need for autonomy and resilience in case a weather event or a power outage threatens to shut down operations.

HPE GreenLake’s consumption-based as-a-service model ensures that organizations can more effectively manage costs with pay-per-use predictability, also providing access to buffer capacity to ensure ease of scalability. This means that organizations don’t have to foot the bill to build out costly IT infrastructure at each edge location but can, rather, contract for capabilities according to specific business needs. HPE also manages the day-to-day responsibilities associated with each environment, ensuring robust security and systems performance while creating opportunity for internal IT organizations to focus on higher-value activities.

As benefits of edge computing get multiplied across processes and locations, the advantages are clear. For example, an additional monthly increase in bottom-line profits of $2,000 per location per month is easily obtained by a per-location HPE GreenLake compute service at, say, $800 per location per month. The net profit, then, is $1,200. When that is multiplied across 1,000 locations, the result is an aggregated profit of an additional $1.2 million per month — or $14.4 million per year. Small positive changes across a distributed enterprise quickly multiply, and tangible results are now within reach.  

As companies build out their edge capabilities and sow the seeds to benefit from a multiplier effect, they should remember to:

• Evaluate what decisions can benefit from being made and acted upon in real time as well as what data is critical to delivering on those insights so the edge environments can be built out accordingly

• Consider scalability — how many sites could benefit from a similar setup and how hard it will be to deploy and operate those distributed environments

• Identify the success factors that lead to revenue gains or cost reductions in a specific edge site and replicate that setup and those workflows at other sites

In the end, the multiplier effect is all about maximizing the potential of edge computing to achieve more efficient operations and maximize overall business success. “We’re in the middle of shifting from an older way of doing things to a new and exciting way of doing things,” Vilfort says. “At HPE we are helping customers find a better way to use distributed technology in their distributed sites to enable their distributed enterprise to run more efficiently.”

For more information, click here.