Edge computing from the front line: Application developer case studies

Edge computing is real. Developers are actively exploring new edges: on‑premises and telecom edge.

Read the Report

What’s driving the need for edge computing?Latency and heavy I/O dominate


Selected latency as the most important reason for pursuing edge computing


Selected Heavy I/O as the key reason—reducing the amount of data travelling through the network


Selected Hyper Local Grouping as the key reason—coordinating fleets of devices in a local area

However, the reasons are more nuanced for individual application developers to pursue edge computing

Overcome limitations of end devices

“We think that the hardware will not reach the level in the next 20 years that we would need to run our software solely on the device”

– Holo-Light

Enable cloud flexibility while maintaining low latency

“The customers need to be able to configure, change, modify and experience the application by themselves”

– 1000 Realities

Keep data local to comply with privacy concerns

“When you disconnect the device, the model is not stored in any way - it needs to be treated as sort of a hot cache that then gets automatically deleted”

– Arvizio

Ensure consistent service to meet regulatory requirements

“Edge can help us achieve the consistent levels of ultra-low latency that regulatory bodies, such as the Federal Aviation Administration, require”

– Atrius

Increase scalability by reducing time to deploy

“For a start-up, it’s mandatory to offer something scalable. And going to set up on the client premise every time takes a lot of resources”

– Holo-Light

Telcos must consider the specific requirements of application developers as they plan edge deployments

Orchestration and management

  • Have standardised orchestration and management of workloads
  • Be able to dynamically move capacity so that it is not wasted
  • Be able to ensure that capacity is available at the nearest edge location to the end-user

Specialised Hardware

  • Provide GPUs at the edge for image processing
  • Provide virtual machines with specific operating systems
  • Ensure fast SSD storage is available

Edge cloud availability

  • Deploy more edge compute locations for increased testing and commercial usage
  • Develop a universal infrastructure that others can tie into