Edge Gateway Workload Offloading for Latency and Energy Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Compute devices often face limitations in executing applications that require specific resources, such as low latency or high performance, due to insufficient local processing power or energy constraints, leading to inability to meet application parameters.
Innovation Solution
A system that utilizes an edge gateway device to offload workloads to edge resources by determining the most suitable set of edge resources based on brokering logic provided by service providers, considering parameters like latency, cost, and service level agreements, to execute tasks efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a compute device uses local resources (general purpose processor and/or accelerator devices) to execute an application, then the device can maintain independence and local processing, but it cannot satisfy latency parameters or energy constraints when local processing power is insufficient
Solution Approach 1:
The patent introduces an edge gateway device as an intermediary between the client compute device and edge resources. The gateway receives offload requests, determines appropriate edge resources based on brokering logic, and manages the offloading process. This mediator enables compute devices to access external processing power without directly managing complex resource allocation, resolving the contradiction between meeting application parameters and maintaining device simplicity.
Solution Approach 2:
The edge gateway device serves multiple functions: it acts as a proxy for offload requests, a broker for matching compute devices with appropriate edge resources, and a manager for monitoring and controlling offloaded workloads. This multi-functional approach allows a single device to handle various aspects of distributed computing, reducing the need for each compute device to have specialized components for each function.
2Productivity
If a compute device offloads workloads to edge resources, then it can access additional processing power and meet application parameters, but it introduces dependency on external resources and network communication
Solution Approach 1:
The system implements self-service mechanisms where the edge gateway automatically determines appropriate edge resources based on brokering logic without requiring manual intervention. The brokering logic autonomously matches compute devices with suitable edge resources based on application requirements, reducing the operational burden on users and simplifying resource management despite the increased productivity capabilities.
Solution Approach 2:
The patent incorporates feedback mechanisms where the edge gateway monitors the status of offloaded workloads and the performance of edge resources. This feedback enables dynamic adjustment of resource allocation and workload distribution, allowing the system to optimize productivity while automatically managing the complexity of external resource coordination without requiring constant manual intervention.
3Speed
If a compute device uses an accelerator device to execute operations faster, then it can meet latency requirements, but it depletes battery energy faster
Solution Approach 1:
The patent extracts the high-energy-consumption accelerator device from the client compute device and relocates it to edge resources. This extraction allows the client device to maintain low power consumption while still accessing fast processing capabilities when needed. The accelerator functionality is removed from the portable device and placed in the external edge infrastructure, resolving the contradiction between speed and energy usage.
Solution Approach 2:
The solution moves the processing capability from a spatial dimension (local to the device) to a network dimension (remote edge resources). Instead of having the accelerator physically attached to or integrated into the client device, the system accesses acceleration capabilities through the network infrastructure. This dimensional shift enables fast processing without the energy cost of local hardware operation.
4Productivity
If a compute device has a fast general purpose processor, then it can execute applications quickly, but it increases device cost and power consumption
Solution Approach 1:
The patent employs a strategy where the client compute device uses a modest, energy-efficient processor for routine tasks and only temporarily accesses high-performance processing capabilities when specifically needed through workload offloading. Instead of continuously operating a fast processor, the system selectively engages edge resources for computationally intensive tasks, reducing average power consumption while maintaining the ability to achieve high productivity when required.
Data Source
AI summary
Technologies for determining a set of edge resources to offload a workload from a client compute device based on a brokering logic provided by a service provider include a device that includes circuitry that is in communication with edge resources. The circuitry is to receive a brokering logic from a service provider receive a request from a client compute device, wherein the request includes a function to be used to execute the request and one or more parameters associated with the client compute device, determine the one or more parameters, select, as a function of the one or more parameters and the brokering logic, a physical implementation to perform the function, wherein the physical implementation indicates a set of edge resources and a performance level for each edge resource of the set of edge resources, and perform, in response to a selection of the physical implementation, the request using the set of edge resources associated with the physical implementation.