Failover Time Measurement for Server Load Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In processing distribution systems combining auto-scaling and automatic failover technologies, operational servers fail to distinguish between load increases due to volume changes or server failures, leading to unnecessary resource utilization and activation of auto-scaling.
Innovation Solution
An information processing device that monitors operational servers, determines the appropriate server for failover based on the number and load, and measures failover time to prevent unnecessary auto-scaling, thereby optimizing resource utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If operational servers monitor load independently, then each server can detect its own load increase, but it cannot distinguish whether the load increase is due to volume change or server failure
Solution Approach 1:
The patent combines the monitoring functions of operational servers with a dedicated management server. The management server aggregates load information from multiple operational servers and centrally determines whether load increases are due to volume changes or server failures, resolving the information loss problem while maintaining measurement precision at the server level.
Solution Approach 2:
The management server acts as an intermediary between operational servers and the auto-scaling system. It receives load information from operational servers, analyzes the cause of load increases, and provides authoritative determination to prevent unnecessary auto-scaling activation, thus preserving information that individual servers cannot obtain.
2Reliability
If auto-scaling is activated on every load increase, then the system responds quickly to potential failures, but unnecessary servers are created when load increase is due to volume change
Solution Approach 1:
The system implements feedback by having operational servers report load information to the management server, which then provides feedback on whether the load increase is due to volume change or server failure. This feedback loop prevents unnecessary auto-scaling activation while maintaining reliable failure detection, thus avoiding resource wastage.
Solution Approach 2:
The management server performs preliminary analysis of load information before triggering auto-scaling. By determining the cause of load increase in advance, the system avoids unnecessary server creation while ensuring rapid response to actual failures, optimizing both reliability and resource efficiency.
3Extent of automation
If management server performs automatic failover, then server failures are handled automatically, but the management server cannot recognize when operational servers activate auto-scaling
Solution Approach 1:
The management server receives feedback information from operational servers about their load status and auto-scaling decisions. This feedback mechanism enables the management server to recognize when operational servers have activated auto-scaling, preventing conflicting actions while maintaining automatic failover capability.
Solution Approach 2:
The management server performs multiple functions including monitoring operational servers, determining failure causes, controlling auto-scaling, and executing failover. This multi-functionality ensures that the management server has complete visibility into system state while maintaining comprehensive automation capabilities.
Data Source
AI summary
An information processing includes a processor and monitors a plurality of operational servers to which processing is allocated. The processor determines an operational server on which failover will be performed in a failover test from among the plurality of operational servers in accordance with a number of the plurality of operational servers and a load, when a condition under which the failover test is conducted is satisfied, and issues a request to measure a failover time of the failover test that is conducted on the determined operational server.