Dynamic Application Object Placement Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional application schedulers fail to dynamically optimize the placement of application objects in distributed computing environments during runtime, leading to suboptimal resource utilization and performance, as they do not consider objective functions like time consumption and energy efficiency, and only ensure resource availability at deployment without runtime migration.
Innovation Solution
A method and system that generate application and resource topology models based on metrics to recommend optimal placement of application objects across network nodes, utilizing execution time, data size, latency, and bandwidth matrices to dynamically adjust object placement and migrate objects to improve performance and energy efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Duration of action of moving object
If one application object is moved from a mobile phone to an edge server, then the computing time for that object is reduced, but the data transmission time between the moved application object and a neighboring application object increases
Solution Approach 1:
The system evaluates different placement locations (mobile phone, edge server, cloud data center) with different local characteristics and selects the optimal location for each application object based on its specific requirements and interactions with other objects, rather than using a uniform placement strategy
Solution Approach 2:
The system dynamically adjusts application object placement during runtime based on changing conditions such as network status, resource availability, and performance metrics, transitioning from static deployment-time placement to dynamic runtime optimization
2Productivity
If application objects are placed to optimize individual component performance, then the performance of that component improves, but the overall network system performance deteriorates
Solution Approach 1:
The system merges the optimization of individual application objects with the optimization of the overall network system by using a unified objective function that considers both local component performance and global system performance, ensuring that local optimizations do not harm overall system reliability
Solution Approach 2:
The system continuously monitors performance metrics and uses feedback loops to adjust placement decisions, ensuring that optimizations at the component level do not degrade overall system performance by detecting and responding to performance changes in real-time
3Ease of manufacture
If conventional application schedulers only ensure resource availability at deployment, then resource allocation is simple, but resource utilization efficiency is suboptimal
Solution Approach 1:
The system enables application objects to be automatically migrated during runtime based on performance metrics and objective functions without requiring complex manual resource allocation planning, allowing the system to self-optimize resource utilization while maintaining simplicity in resource management
Solution Approach 2:
The system performs preliminary evaluation of placement options using objective functions and performance metrics before making migration decisions, ensuring that resource utilization is optimized in advance rather than reacting to performance degradation
Data Source
AI summary
A computer-implemented method for placement of a plurality of application objects of an application within a network architecture is disclosed. The method includes generating during runtime of the application, an application topology model for the application, based on application metrics for the plurality of application objects. A resource topology model of a plurality of network nodes within the network architecture is generated based on resource metrics for the network nodes. A recommendation is generated for migrating an application object of the plurality of application objects to a network node of the plurality of network nodes using the application topology model and the resource topology model, the recommendation identifying the application object and the network node. The application object is migrated to the network node identified by the recommendation.