Adaptive Demand-Driven Load Balancing for Business Applications
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing load balancing mechanisms in business applications are inefficient due to static workload distribution, which does not account for varying computational complexities of work packages, leading to uneven processor utilization and potential bottlenecks, especially when processing jobs with different complexities across multiple processors.
Innovation Solution
Implementing an adaptive demand-driven load balancing system that uses an application-dependent algorithm to distribute a subset of the workload initially, based on estimated processing times and complexities, and dynamically assigns remaining work items as processors become available, thereby optimizing processor utilization without requiring extensive domain-specific knowledge or additional processing resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If static load balancing is used to distribute workload evenly, then workload distribution is simple, but processor utilization becomes inefficient due to varying computational complexities
Solution Approach 1:
The patent implements dynamic load balancing that adapts to varying computational complexities of work packages. Instead of static equal distribution, the system dynamically adjusts workload allocation based on the actual complexity of each package, allowing processors to receive varying amounts of work that better matches their processing needs and capabilities, thereby improving processor utilization efficiency.
Solution Approach 2:
The system changes the parameter of workload distribution from fixed equal shares to variable shares based on computational complexity. By introducing complexity as a weighting parameter, the load balancer adjusts the distribution strategy to account for differences in processing requirements, transforming the simple equal distribution model into a complex adaptive model that optimizes processor utilization.
2Productivity
If load balancer analyzes and estimates processing requirements for each work package, then processor utilization improves, but processing time increases and bottleneck occurs
Solution Approach 1:
The patent applies preliminary action by pre-defining complexity thresholds and classification rules for work packages before the load balancing process begins. The system pre-establishes criteria for identifying different complexity levels, allowing work packages to be rapidly classified without requiring real-time complex analysis, thus improving processor utilization while minimizing load balancer processing time.
Solution Approach 2:
The system uses simple, lightweight complexity indicators or tags attached to work packages rather than performing expensive comprehensive analysis. These simple markers provide sufficient information for load balancing decisions without requiring extensive processing resources, enabling efficient workload distribution without creating a bottleneck at the load balancer.
3Ease of operation
If work packages of same size are distributed, then distribution is straightforward, but work processes complete tasks at different rates leading to idle processors
Solution Approach 1:
The patent applies local quality by tailoring the workload distribution to the specific characteristics of each work package rather than treating all packages uniformly. Each work package receives appropriate attention based on its individual complexity, ensuring that processors receive work that matches their processing capabilities and that all work is completed in a coordinated manner, improving task completion consistency across processors.
Data Source
AI summary
The present disclosure involves systems, software, and computer implemented methods for providing adaptive demand-driven load balancing for processing jobs in business applications. One process includes operations for identifying a workload for distribution among a plurality of work processes. A subset of the workload is assigned to a plurality of work processes for processing of the subset of the workload based on an application-dependent algorithm. An indication of availability is received from one of the plurality of work processes, and a new subset of the workload is assigned to the work process.