Predictive Routing for Load Balancing

Solution Overview

Problem

Conventional load balancing systems face challenges in accurately routing requests due to stale measurement data, which can lead to inefficient distribution of network traffic, and frequent measurement updates increase the load on destination servers.

Innovation Solution

The system predicts future measurement data using past data and routes requests based on predicted values, incorporating factors like traffic load, CPU load, response time, and proximity to ensure optimal routing without overburdening the servers.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If measurements are taken at more frequent time intervals to reduce staleness, then measurement accuracy is improved, but the load on destination servers increases and computational cost increases

Engineering Contradiction:

Improvemeasurement accuracyVSAvoidserver load

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The system performs prediction calculations in advance based on historical measurement data, generating predicted future states before actual routing decisions are needed. This allows the load balancer to have current accuracy information without performing frequent actual measurements, thereby reducing the load on destination servers while maintaining measurement precision.

Inventive Principle:

Principle #10Preliminary action

2Measurement precision

If measurements are taken at more frequent time intervals to reduce staleness, then measurement accuracy is improved, but computational expense increases

Engineering Contradiction:

Improvemeasurement accuracyVSAvoidcomputational expense

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

Instead of performing actual frequent measurements which consume computational resources, the system creates a predictive model that copies the essential patterns from historical data. This model can then generate accurate predictions without requiring repeated expensive measurement operations, thus reducing computational expense while maintaining measurement accuracy.

Inventive Principle:

Principle #26Copying

3Measurement precision

If the load balancer uses current measurement data to route requests, then routing accuracy is improved, but network delays cause the data to become stale

Engineering Contradiction:

Improverouting accuracyVSAvoiddata staleness

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system performs prediction calculations in advance based on historical measurement data, generating predicted future states before actual routing decisions are needed. This allows the load balancer to have current accuracy information without performing frequent actual measurements, thereby reducing the load on destination servers while maintaining measurement precision.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS10652318B2Systems and methods for load balancing using predictive routing

Publication Date: 2020.05.12 VERISIGN INC

AI summary

Systems and methods are disclosed for routing requests for information based on predictive data. The systems and methods may receive measurement data indicative of states of each of a plurality of destination servers, and generate predicted measurement data values for each of the plurality of destination servers based on the retrieved measurement data. The predicted measurement data values may represent predicted states of each of the destination servers at a time later than a time corresponding to the received measurement data. The systems and methods may also receive requests for information from a client computer, and route the received requests for information to one of the plurality of destination servers based on the predicted measurement data value.