Transaction System Record-Level Exclusion Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing transaction systems that support transactions on multiple Key Value (KV) data in a Key Value Store (KVS) face decreased concurrent executability and throughput due to exclusion violations when transactions are grouped into a single unit, making it difficult to scale out and maintain high performance.
Innovation Solution
A transaction system that includes a Key Value data storage unit, a Write Ahead Log (WAL) storage unit for storing update histories, and processing units to determine exclusion violations and record updates, allowing transactions to succeed only if no exclusion violations are detected, thereby updating the KV data storage unit based on the WAL.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple KV data are grouped into one KV data to support transactions on multiple KV data, then transaction support on multiple KV data is enabled, but concurrent executability of transactions decreases due to exclusion violations
Solution Approach 1:
The patent segments the exclusion control unit from the data grouping unit. Multiple KV data are grouped into one KV data for transaction support, but exclusion control is performed on individual original KV data items rather than on the grouped KV data. This segmentation allows transactions to proceed concurrently on different original KV data without unnecessary exclusion violations, while still maintaining transactional integrity across multiple KV data through the grouping mechanism.
2Productivity
If pessimistic exclusion is applied to perform exclusion with each KV data as exclusion unit, then concurrent executability improves, but scaling out becomes difficult and KV data advantage is lost
Solution Approach 1:
The patent segments exclusion control into fine-grained units (individual original KV data) while keeping the storage structure coarse-grained (grouped KV data). This allows optimistic exclusion to be applied at the KV data level without requiring pessimistic locking mechanisms, maintaining the scalability and simplicity of KV data-based storage while achieving high concurrent executability through record-level exclusion control.
Solution Approach 2:
The patent introduces an intermediary mechanism (the grouping structure) that bridges between individual KV data items and transactional operations. The grouped KV data serves as an intermediary that maintains references to original KV data, allowing exclusion control to be applied at the appropriate level without direct interference between concurrent transactions, thus enabling both high concurrency and scalability.
3Device complexity
If optimistic exclusion is used on grouped KV data, then scaling out is facilitated, but unnecessary exclusion violations occur reducing throughput
Solution Approach 1:
The patent segments the exclusion control granularity from the data grouping granularity. By controlling exclusion at the level of individual original KV data rather than at the grouped KV data level, the system avoids unnecessary exclusion violations that would occur when treating grouped data as a single exclusion unit. This segmentation maintains the scalability benefits of optimistic exclusion while improving throughput by allowing more transactions to proceed concurrently without false exclusions.
Data Source
AI summary
A KV data storage part stores a plurality of key value data each being a pair of key and value. The value contains a record of tabular data. The key contains part of column values of the record. A WAL storage part stores an update history of a data group. The data group includes a plurality of KV data referred to and updated by one transaction. The update history has information of a data manipulation instruction in the transaction and the record changed through processing according to the data manipulation instruction. A transaction processor, in commit of the transaction, determines the presence or absence of exclusion violation per record on a basis of the update history stored in the WAL storage part and, if exclusion violation is absent, makes the transaction succeed and records the update history relating to the transaction into the WAL storage part.