Redundant Coordinator Architecture for Distributed Transaction Consistency

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The two-phase commit technique in distributed transactions suffers from single point of failure issues, communication delays, and heuristic inconsistencies, particularly due to the transaction coordinator being a single point of failure and the branch portion being a single point of failure, which leads to stagnation in business operations.

Innovation Solution

A distributed transaction control system with redundant coordinator apparatuses that include a determination part, a distributed consensus processing part, and a settlement part, where the determination part waits for input votes, replaces timed-out votes, and makes decisions on commit or rollback based on vote outcomes, using distributed consensus to decide the transaction outcome and notify participant apparatuses.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If a single transaction coordinator is used in two-phase commit, then the system structure is simple, but the coordinator becomes a single point of failure causing business operation stagnation

Engineering Contradiction:

Improvesystem structureVSAvoidsingle point of failure

Core Design Contradiction:

Device complexityVSReliability

Solution Approach 1:

The patent applies local quality by differentiating the roles and capabilities of different coordinator nodes. Primary coordinators have full transaction coordination capabilities while backup coordinators have limited capabilities, creating localized functional differences that improve overall system reliability without requiring complete functional redundancy across all nodes.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The patent implements beforehand cushioning by pre-establishing backup coordinator nodes that can take over if the primary coordinator fails. This proactive preparation of redundant resources before failure occurs ensures continuous transaction coordination capability and prevents business operation stagnation.

Inventive Principle:

Principle #11Beforehand cushioning (Prior cushioning)

2Reliability

If distributed consensus is used to make the transaction coordinator redundant, then the single point of failure problem is solved, but the commit process complexity increases and compatibility with XA is lost

Engineering Contradiction:

Improvesingle point of failureVSAvoidcommit process

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent segments the coordinator functionality into distinct primary and backup coordinator roles with different responsibilities. This segmentation allows the system to achieve redundancy and fault tolerance while maintaining a simpler, more XA-compatible commit process compared to full distributed consensus implementations.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent implements dynamic role assignment where coordinators can transition between primary and backup states. This dynamic flexibility allows the system to adapt to failure scenarios while maintaining operational simplicity and XA compatibility during normal operation, only activating full redundancy mechanisms when needed.

Inventive Principle:

Principle #15Dynamics

3Adaptability or versatility

If two-phase commit is performed along distributed application topology, then the transaction coordination follows the application structure, but the branch portion becomes a single point of failure and communication time increases

Engineering Contradiction:

Improvetopology followingVSAvoidbranch portion single point of failure

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The patent introduces intermediary coordinator nodes that mediate between application branches and the central coordination logic. These intermediary coordinators prevent branch portions from becoming single points of failure by distributing coordination capabilities while still allowing the system to follow application topology for transaction management.

Inventive Principle:

Principle #24Intermediary (Mediator)

4Reliability

If communication delay occurs in two-phase commit, then the transaction coordination completes eventually, but heuristic inconsistency occurs and business operations stagnate

Engineering Contradiction:

Improvetransaction consistencyVSAvoidcommunication delay

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent applies preliminary action by having backup coordinators pre-position themselves and maintain readiness to take over coordination responsibilities. This preliminary preparation reduces the time penalty of failures and prevents heuristic inconsistency by ensuring a coordinated response is already in place, minimizing business operation stagnation.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS12105603B2Distributed transaction control system and distributed transaction control method

Publication Date: 2024.10.01 HITACHI LTD

AI summary

A distributed transaction control system includes redundant coordinator apparatuses, wherein determination part replaces input votes to votes of preparation failure when input votes from the participant apparatus time out, makes a determination of commit when all input votes are votes of preparation success, makes a determination of rollback when the input votes include at least one vote of preparation failure, and inputs a content of the determination into the distributed consensus processing part, the distributed consensus processing part decides on one of the contents of the determination that are inputs from the determination parts, by distributed consensus, and settlement part notifies an output of the distributed consensus processing part to the participant apparatuses as a settlement result.