Despite heavy adoption of cloud services, chief information officers are finding that data-intensive applications in areas like the internet of things, artificial intelligence, automation, augmented reality and virtual reality are not always a fit for centralized cloud models. The need for better performance and real-time analytics is pushing many organizations to the edge.
Edge computing is a distributed model in which compute activities are performed outside of a centralized data center. IDC anticipates worldwide spending on edge computing to hit $176 billion in 2022, a 14.8% increase over 2021. The largest investments cut across a variety of use cases, including manufacturing operations, production asset management, smart grids, omnichannel operations, freight monitoring, and public safety and emergency response applications.
Edge deployments complement cloud architectures by tackling latency issues associated with running compute-intensive applications in the cloud. The volume of data required for real-time insights also makes it cost-prohibitive to migrate and store everything on a centralized cloud platform. In one IDC survey, 73% of executives called edge computing a strategic investment and three-quarters of those respondents said they expect less than 5 millisecond latency for key applications — a benchmark not feasible with centralized or cloud computing models.
“Latency is a major driver because of the processing power and need to make quick decisions sits at the edge,” says Amr Ahmed, EY Americas Infrastructure and Service Resiliency Leader. “It could be a manufacturing site, an oil rig, a campus building or a data center — you can’t send data back to the cloud when submicrosecond decisions are required.”
Edge at work
Edge computing is gaining a foothold across industries. Here is a snapshot of how three key sectors are capitalizing on its capabilities:
- Manufacturing: Taking frequent, high-resolution pictures of a high-pressure pipe to monitor for cracks creates a massive amount of data, which can clog the bandwidth if sent back to a central data center. Processing this data at the edge reduces bandwidth and allows for a rapid response if a defect is detected. Data specific to the defect can then be sent back to the cloud for further analysis and to share with other facilities that have similar equipment to proactively address defects before they impact production.
- Health care: Health providers are deploying artificial intelligence for genomic sequencing to identify genetic variants that impact a patient’s health. The process creates terabytes of data, but once the sequence is identified, the data no longer has value. Collecting and processing the data locally, then transmitting only the genomic sequence back to a central cloud service or server saves time and bandwidth.
- Retail: On-site point-of-sale systems use scanners to process data. Sending those bar codes to a central location to look up a price would significantly slow transactions. Edge capabilities within individual supermarkets perform price lookups locally, elevating the customer experience by reducing wait times.
Chief information officers should look at edge capabilities as a complementary pillar of the IT estate, driven by the business case. Organizations need to understand the overall edge architecture, which includes operations, connectivity, security, and the interplay with private and public clouds, as well as on-premises resources.
Integration across the IT estate remains key to a successful edge deployment. “Adopt the edge technology that integrates seamlessly with your cloud strategy, and have a plan to integrate and manage all edge servers and data coming from them,” explains Mounir Elmously, Senior Manager, Technology Consulting, Ernst & Young LLP. “You want to avoid creating separate silos.”
Learn more about actions chief information officers can take today to digitally transform their business.
The views expressed by the author are not necessarily those of Ernst & Young LLP or other members of the global EY organization.