Online Timestamp-based Transactional Isolation Checking of Database Systems (Extended Version)
Hexu Li, Hengfeng Wei, Hongrong Ouyang, Yuxing Chen, Na Yang, Ruohao Zhang, Anqun Pan·April 02, 2025
Summary
CHRONOS, AION, and AION-SER streamline transactional isolation checks. CHRONOS simplifies extensive history management. AION and AION-SER excel in online scenarios, with AION focusing on Snapshot Isolation and AION-SER on Serializability. Experiments highlight high throughput and practical application. Aion vs. Cobra comparisons show Aion's performance drops with GC, while Cobra maintains stable but capped at 3K TPS. AION detects EXT violations with stable performance influenced by asynchrony, with false positives/negatives resolved within 10ms to 1.5 seconds. Higher time thresholds mitigate impacts. AION efficiently detects violations, with checking speeds comparable to histories without specific issues. TPCC workloads are examined for offline checking, and AION's online checking throughput for non-conforming histories is presented.
Introduction
Background
Context of transactional isolation checks
Objective
Streamlining and simplifying transactional isolation checks
CHRONOS: Simplifying Extensive History Management
Functionality
Overview of CHRONOS's role
Implementation
How CHRONOS simplifies history management
AION and AION-SER: Online Scenarios
AION: Snapshot Isolation
Focus and benefits of Snapshot Isolation
AION-SER: Serializability
Focus and benefits of Serializability
Comparison
Distinction between AION and AION-SER
Experiments and High Throughput
Throughput Analysis
Highlighting high throughput capabilities
Practical Application
Real-world application and implications
AION vs. Cobra Performance Comparison
Performance Metrics
Key performance indicators (KPIs) for comparison
GC Impact on AION
Performance drop with Garbage Collection (GC)
Cobra's Performance
Stable but capped throughput at 3K TPS
AION's Detection of EXT Violations
Performance under Asynchrony
Influence of asynchrony on performance
False Positives/Negatives
Resolution time for false positives/negatives
Time Thresholds
Impact of higher time thresholds on performance
Efficient Violation Detection
Checking Speeds
Comparison with histories without specific issues
TPCC Workloads
Examination of TPCC workloads for offline checking
AION's Online Checking Throughput
Throughput for Non-Conforming Histories
Presentation of AION's online checking throughput
Conclusion
Summary of Key Findings
Future Directions
Potential areas for further research and development
Basic info
papers
databases
Advanced features
Insights
What are the key differences in performance between AION and Cobra, particularly in terms of throughput and garbage collection impact?
How do CHRONOS, AION, and AION-SER streamline transactional isolation checks?
What are the throughput capabilities of AION's online checking for non-conforming histories in TPCC workloads?
How does AION handle EXT violations and what are the implications of time thresholds on its performance?