Managing Transactions in Snapshot Isolation with Adaptive Timeouts

Authors: Vipul Kumar Bondugula

DOI: https://doi.org/10.37082/IJIRMPS.v9.i4.232466

Short DOI: https://doi.org/g9hnp3

Country: India

Full-text Research PDF File:   View   |   Download

Abstract: Transaction management in databases plays a critical role in ensuring consistency, isolation, and correctness during concurrent operations. Among various concurrency control mechanisms, Snapshot Isolation (SI) is widely adopted for its balance between performance and isolation. SI allows transactions to execute on a consistent snapshot of the database, thus avoiding locking overhead and improving throughput. However, one major challenge under SI is handling transaction conflicts, particularly under fixed timeout strategies. Fixed timeouts set a predefined waiting period for a transaction to complete or abort, regardless of workload or system conditions. When contention is high or workloads vary dynamically, fixed timeouts may not provide enough flexibility, causing premature aborts or unnecessary waiting, both of which negatively affect system throughput. Under fixed timeout implementations in SI, transactions are often retried due to expired wait times, especially in high-contention environments. These retries significantly increase the total number of operations and reduce efficiency. As transaction volume and data contention grow, fixed timeouts result in more frequent retries, wasting computation and increasing latencyThese methods aim to reduce unnecessary retries by giving transactions sufficient time to complete when the system is under moderate load or by preemptively aborting transactions under high contention. By learning from previous executions or estimating expected completion windows, dynamic timeouts allow better resource utilization and reduce overhead. When compared numerically, dynamic timeout systems showed lower retry counts across all node configurations, offering a more scalable and efficient solution. Fixed timeout strategies often fail to adapt to system variability, leading to higher retry counts and reduced efficiency. Dynamic timeout approaches provide a more intelligent, context-aware alternative to manage transactions effectively. This paper addresses the retry count of fixed time out process by using the dynamic timeout process.

Keywords:

Paper Id: 232466

Published On: 2021-08-12

Published In: Volume 9, Issue 4, July-August 2021

Share this