Load Balancer Score for Server Utilization and Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current load balancing techniques are inadequate for efficiently distributing client processing jobs across a large number of servers, particularly in mobile communication systems, as they do not effectively account for both server load and geographical separation, leading to potential overloading and performance issues.
Innovation Solution
A load balancer system that generates a load balancing score for each server based on client ID, server ID, processing load metric, and separation metric, preferring servers with lower separation but avoiding overload by considering network latency and geographic distance, and assigns jobs to servers with the lowest score.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If load balancing assigns client jobs to servers with lowest load (least load approach), then server utilization is improved, but geographical separation and network latency are not considered leading to potential performance degradation
Solution Approach 1:
The patent changes the load balancing parameter from simple load metrics to a composite score that incorporates both server load and geographical separation metrics. This allows the system to simultaneously optimize for server utilization and network latency by selecting servers based on the composite score rather than load alone.
Solution Approach 2:
The patent creates a composite load balancing score that combines multiple factors (server load, geographical distance, network latency) into a single metric. This composite approach allows the system to balance multiple competing objectives simultaneously, resolving the contradiction between utilization and latency.
2Loss of time
If load balancing prefers servers with minimal geographical separation, then network latency is reduced, but servers may become overloaded
Solution Approach 1:
The patent modifies the selection criterion from purely geographical proximity to a composite score that balances both proximity and server load. This ensures that while geographical separation is considered, it does not lead to server overload by incorporating load metrics into the scoring mechanism.
3Productivity
If round-robin load balancing is used to distribute jobs evenly, then server utilization is improved, but geographical separation and network latency are not considered
Solution Approach 1:
The patent replaces the simple sequential assignment of round-robin with a score-based selection mechanism that incorporates both load and geographical metrics. This maintains even distribution while optimizing for response time by considering multiple parameters in the job assignment decision.
4Stability of the object's composition
If hash routing is used to assign jobs based on source-destination addresses, then job routing consistency is improved, but load balancing and geographical considerations are lost
Solution Approach 1:
The patent merges the consistency aspect of hash routing with the load balancing and geographical considerations of score-based selection. By incorporating hash values into the composite score calculation, the system maintains routing consistency while simultaneously optimizing for load distribution and geographical separation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A communications system may include a plurality of client devices configured to generate client processing jobs, each having a respective unique client ID associated therewith. A plurality of geographically spaced apart servers may be configured to process the client processing jobs, each having a respective unique server ID associated therewith. A load balancer may be configured to receive the client processing jobs from the plurality of client devices, and for each client processing job, generate a respective load balancing score for each server based upon the client ID of the client device requesting the client processing job, the server ID of each server, a processing load metric for each server, and a separation metric corresponding to a separation between the requesting client device and each server. The load balancer may also be configured to assign each client processing job to a respective server based upon the load balancing scores.