Client Replication Service Mining Engine for Database Rolling Upgrades

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

During a database upgrade, the transition from physical to logical replication in a rolling upgrade process disrupts log mining operations, leading to incomplete redo log processing and loss of session information, resulting in service interruptions and inconsistent records due to the lack of support for mining features in logical standby databases.

Innovation Solution

Implementing a downstream mode for the client replication service, where foreign redo logs are transported and registered on the new primary DBMS, allowing the mining engine to resume processing from the original primary DBMS's unmined logs, ensuring seamless log mining and maintaining mining features support.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Duration of action of stationary object

If physical standby is converted to logical standby during rolling upgrade, then database upgrade continuity is maintained, but log mining operations are disrupted and session information is lost

Engineering Contradiction:
Improvedatabase upgrade continuityVSAvoidlog mining operations
Core Design Contradiction:
Duration of action of stationary objectVSReliability

Solution Approach 1:

The patent introduces an intermediary mechanism that captures and preserves mining session information and dictionary data during the transition from physical to logical standby. This intermediary storage allows the mining operations to resume after the upgrade without losing critical session state, thus maintaining reliability while enabling upgrade continuity.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent performs preliminary actions by saving the mining session information and dictionary to a persistent location before the physical-to-logical standby conversion occurs. This preliminary preservation of data ensures that when the conversion is complete, the mining operations can immediately resume from the saved state rather than starting over or failing.

Inventive Principle:
Principle #10Preliminary action

2Device complexity

If mining session information is not preserved during role switch, then system complexity is reduced, but service interruption occurs and records become inconsistent

Engineering Contradiction:
Improvesession information managementVSAvoidservice continuity
Core Design Contradiction:
Device complexityVSProductivity

Solution Approach 1:

The patent creates a copy of the mining session information and dictionary data and stores it in a persistent location that survives the role switch. This copying mechanism allows the session state to be preserved without adding significant complexity to the system architecture, while ensuring service continuity and record consistency.

Inventive Principle:
Principle #26Copying

3Adaptability or versatility

If logical replication is used instead of physical replication, then heterogeneous database support is enabled, but mining features are not supported and data consistency is compromised

Engineering Contradiction:
Improveheterogeneous database supportVSAvoidmining features support
Core Design Contradiction:
Adaptability or versatilityVSReliability

Solution Approach 1:

The patent introduces an intermediary persistent storage mechanism that bridges the gap between logical replication's heterogeneity support and mining features' reliability requirements. By storing session information externally, the system can use logical replication for flexibility while maintaining mining operation integrity through the intermediary storage layer.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentUS11481367B2Redo mining during rolling upgrade
Publication Date: 2022.10.25 ORACLE INT CORP
  • US11481367B2 patent drawing
  • US11481367B2 patent drawing
  • US11481367B2 patent drawing

AI summary

In some embodiments, a client replication service manages log mining during a rolling upgrade, in which the primary role is switched over from an original primary DBMS to a new primary DBMS having updated software. The client replication service or specifically a mining engine maintains an original session on the original primary DBMS to mine redo logs associated with the original primary DBMS until the original primary DBMS is not a primary DBMS anymore. At that time, if the mining of those redo logs is not complete, the client replication service further starts a downstream session on the new primary DBMS to mine copies of the unmined redo logs.