In many cloud computing environments, cloud-native applications can be decoupled from the underlying hardware that operates such applications. Presented herein are techniques through which capabilities and metrics collected from a virtual/physical compute resource can be used to identify the presence of different accelerators and their capabilities in real time. The metrics that can be collected from within an application can be extended to include execution context awareness, which may be used to suggest a resource profile for the application. Further, holistic visualization of various metrics can be provided in some instances based on the accelerator from where a service is being executed.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.