Abstract:OLTP has stringent performance requirements defined by Service Level Agreements. Transaction response time is used to determine the maximum throughout in benchmarks. Capacity planning tools for OLTP performance are based on queueing network models for hardware resources and database lock contention has a secondary effect on performance. With ever increasing levels of e-commerce and surges in OLTP traffic we discuss the need for studies of database workloads to develop more realistic lock/latch contention models. Predictive formulas to model increased load leading to thrashing for txns with identical and nonidentical steps are presented. We review concurrency control methods to reduce the level of lock/data conflicts in high contention environments.
What problem does this paper attempt to address?
The main problems that this paper attempts to solve are the concurrency control and high data contention issues in Online Transaction Processing (OLTP) systems. Specifically, the author focuses on the impact of database lock contention on system performance and proposes the need to study more realistic lock / latch contention models to cope with the increasing e - commerce activities and the surge in OLTP traffic.
### Main problems:
1. **The impact of lock contention on OLTP system performance**:
- Traditional queuing network models mainly focus on the contention of hardware resources, ignoring the secondary impact of database lock contention on performance.
- With the increase in e - commerce activities and the surge in OLTP traffic, the existing lock contention models may not be sufficient to accurately describe the actual scenarios.
2. **Concurrency control methods in high - data - contention environments**:
- In a high - data - contention environment, how to reduce lock / data conflicts and improve the system's throughput and response time.
- Research different concurrency control methods, such as Two - Phase Locking (2PL), Optimistic Concurrency Control (OCC), etc., to find a method more suitable for high - contention environments.
3. **Prediction formulas and analysis methods**:
- Propose prediction formulas to model system thrashing caused by load increases, especially in cases where transactions have the same and different steps.
- Evaluate the performance of transaction processing systems through analysis methods, considering the impact of lock contention.
### Solutions:
- **Heterogeneous Data Access Model (HDAM)**: A method for concurrency control and handling high - data - contention.
- **Probability analysis of lock contention and deadlock**: Calculate the probabilities of lock contention and deadlock through mathematical formulas (such as formulas (1), (2), (4), etc.).
- **Comparison of multiple concurrency control methods**: Including methods such as Wait - Die (WD), Wound Wait (WW), No Waiting (NW), Cautious Waiting (CW), Running Priority (RP), etc., to find the method most suitable for high - contention environments.
- **Adaptive methods and feedback mechanisms**: Dynamically adjust the concurrency level according to the current state of the system to prevent the system from entering a thrashing state.
### Conclusions:
The paper emphasizes that in OLTP systems, in addition to the contention of hardware resources, the contention of software resources (such as lock and latch contention) is equally important. Through in - depth research and analysis of lock contention, the performance bottlenecks of the system can be better understood, and effective solutions can be proposed to improve the system's throughput and response time.
Hope this information can help you understand the core problems and solution ideas of this paper. If you have more questions or need further explanations, please feel free to let me know!