Edge Compute Instance Optimization via Local QoS Resource Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Centralized controllers in edge computing networks face increased management traffic and scalability issues, leading to network overhead and single points of failure when managing resource allocation for virtualized network functions (VNFs).
Innovation Solution
Implementing a performance manager as a user space thread within virtualized systems to monitor resource usage, profile performance, and manage resource allocation, reducing reliance on centralized controllers and distributing resource management to execute VNFs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a centralized controller is used to manage resource monitoring and allocation for VMs executing VNFs, then centralized control and management is improved, but network overhead increases and scalability deteriorates
Solution Approach 1:
The patent segments the centralized resource management function by introducing local resource monitors on each compute device. These local monitors handle resource monitoring and allocation decisions autonomously, dividing the previously centralized management task into distributed components. This reduces network traffic for management operations while maintaining effective resource control.
Solution Approach 2:
The patent implements self-service mechanisms where compute devices automatically monitor their own resource usage and make allocation decisions without constant centralized intervention. The local resource monitors on compute devices autonomously manage VM resource allocation based on predefined policies, reducing dependency on the centralized controller and minimizing management traffic.
2Ease of operation
If a centralized controller is used to manage resource monitoring and allocation for VMs executing VNFs, then centralized control is improved, but scalability worsens
Solution Approach 1:
The patent divides the resource management functionality into segmented units distributed across multiple compute devices. Each compute device has its own local resource monitor that can independently manage resources, allowing the system to scale horizontally by simply adding more compute devices without proportionally increasing centralized controller burden.
Solution Approach 2:
The patent transitions from a single-dimensional centralized control model to a multi-dimensional distributed architecture. Management functions operate at multiple levels: local resource monitors on compute devices handle immediate resource decisions, while the centralized orchestrator handles higher-level coordination. This dimensional expansion enables better scalability.
3Ease of operation
If a centralized controller is used to manage resource allocation, then centralized management is improved, but reliability deteriorates due to single point of failure
Solution Approach 1:
The patent segments the management architecture into multiple independent components: local resource monitors on each compute device and a centralized orchestrator. This segmentation creates redundancy, as the failure of the centralized orchestrator does not completely disable resource management, as local monitors can continue to operate autonomously.
Solution Approach 2:
The patent implements self-service capabilities in local resource monitors that allow compute devices to continue managing their own resources even when the centralized controller is unavailable. This autonomous operation ensures system reliability and prevents complete system failure due to centralized controller downtime.
Data Source
AI summary
Technologies for analyzing and optimizing workloads (e.g., virtual network functions) executing on edge resources are disclosed. According to one embodiment disclosed herein, a compute device launches a virtualized system including a virtual network function and a performance manager, the performance manager to monitor a current resource usage of the virtual network function as a function of a performance profile. The compute device determines, in response to a determination that one or more quality-of-service (QoS) requirements is not satisfied, whether one or more resources from the platform are available for satisfying the QoS requirements. The compute device receives, in response to a determination that the one or more resources are available for satisfying the QoS requirements, the one or more resources and updates the performance profile as a function of the received resources.