Whether in a factory or on an oil

[rakhirhif8963]

To put the scale into perspective, McDermott notes that semiconductor electronics used to make up about 5% of the cost of a solution. Today, it's more than 40% of the cost.

rig, AI and other forms of automation increasingly require “proximity” – that is, computing power and other IT infrastructure nearby.

This is consistent with another big trend in edge computing: As the technology matures, the capabilities expand. This is true in terms of hardware and architecture in manufacturing, Satyanathan says. “With advances in AI and edge servers with GPU-centric architectures becoming available, it’s a good idea for manufacturing CIOs to start putting AI applications at the edge,” he says.

This is also true from a software and bolivia mobile database perspective. Management is a key challenge in edge environments in general, and you can’t send a support engineer to every edge node every time something goes wrong. Open source projects like MicroShift help extend mission-critical platforms like Kubernetes to the edge and ensure consistency across them.

3. It's not an all or nothing choice.
Like most other aspects of edge computing strategy—and cloud computing strategies—it’s not necessarily a pro or con decision. This is doubly true when it comes to running AI/ML workloads at the edge.

Red Hat technology evangelist Gordon Huff has made a number of points about the value of taking advantage of established patterns in edge architecture, one of which is this: You can develop centrally and infer locally. “One of the portfolio architectures that is particularly relevant to the edge is the industrial edge, which is applicable to several verticals, including manufacturing,” he says. “It describes the routing of sensor data for two purposes: model development in the central data center and real-time inference in the factory data centers.”