Adaptive Application Deployment for Heterogeneous Edge Clusters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In edge computing scenarios, existing configuration files cannot adaptively deploy applications on diverse hardware types, as they are uniformly managed from the cloud side, failing to meet the varying computing and storage capabilities of heterogeneous devices, such as home gateways and industrial robots with different CPU architectures.
Innovation Solution
An application deployment method that receives a blueprint with sub-blueprints corresponding to specific hardware types and their external capabilities, allowing for adaptive deployment on hardware devices in a cluster to meet the total external capability requirement, and a method to generate such blueprints for microservice platforms to orchestrate and schedule microservices effectively across multiple hardware types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If one configuration file is used to manage applications uniformly from the cloud side, then deployment management is simplified, but the system cannot adapt to heterogeneous hardware types with different computing and storage capabilities at the edge
Solution Approach 1:
The patent segments the configuration file into a hierarchical structure with a global configuration section and multiple hardware-type-specific configuration sections. Each hardware type (e.g., home gateway, industrial robot) has its own dedicated configuration block that defines external capabilities specific to that hardware. This segmentation allows the system to adapt to heterogeneous hardware while maintaining a unified configuration management approach through the global section.
2Productivity
If the system deploys applications on diverse edge devices with varying capabilities, then resource utilization is optimized, but determining the number of application instances and their deployment locations becomes complex
Solution Approach 1:
The patent implements a feedback mechanism where the system automatically collects information about available hardware resources and their external capabilities, then uses this feedback to compute and determine the optimal number of application instances and their deployment locations. The configuration file structure enables the system to read hardware-specific capabilities and automatically make deployment decisions without manual intervention, simplifying the operation while optimizing resource utilization.
3Reliability
If application instances are deployed on specific hardware types based on their capabilities, then external capability requirements are met, but the deployment process becomes more complex
Solution Approach 1:
The patent applies preliminary action by pre-defining all hardware-type-specific configurations and external capabilities in the configuration file before deployment. The system pre-processes this configuration information to automatically determine instance counts and deployment targets. This preliminary preparation ensures that external capability requirements are reliably met while the actual deployment process remains simple, as the complex decisions have already been made during configuration processing.
Data Source
AI summary
Provided is an application deployment method, including: receiving an application blueprint (S1) including at least two different sub-blueprints; and deploying an application instance on a hardware device in a cluster according to the application blueprint and external capabilities provided by all deployed application instances (S2), with a sum of the external capabilities provided by all the deployed application instances being greater than or equal to the total external capability of the application. The present disclosure further provides an application blueprint generation method, a microservice platform, a communication terminal and a computer-readable medium.