Análisis de logs y wait events como metodología estructurada para el diagnóstico del rendimiento en bases de datos relacionales gestionadas
Performance issues in relational databases are often diagnosed reactively through the isolated analysis of metrics, making it difficult to identify their underlying causes. The objective of this study was to measure the effectiveness of a structured methodology based on logs and wait events. A quantitative approach with an applied quasi-experimental design was employed to compare two conditions on the same PostgreSQL 16 instance: a baseline condition without supporting indexes and an optimized condition following a diagnosis-driven intervention. A concurrent workload was generated using `pgbench` on a synthetic dataset comprising 32,050,000 records. Performance Insights, EXPLAIN ANALYZE, and `pg_stat_statements` were used to analyze wait events, execution plans, and query statistics. During the three complete minutes of the baseline test, the average active load was 14.867 sessions, compared with 2 available vCPUs. Following the intervention, throughput increased from 0.43 to 203.48 transactions per second, while mean latency decreased from 18,409 to 39 milliseconds. The findings provided favorable evidence of the methodology's effectiveness under the evaluated conditions, although its applicability to other environments requires further validation.
Authors
- Gabriel Sánchez Bautísta (ORCID: https://orcid.org/0000-0002-9955-8711)
- Angel Enrique Romero Cuevas
Institutions
- Universidad Autónoma del Estado de Hidalgo (MX)
Publication Details
- Journal
- Sciencevolution
- Published
- 2026-11-16
- DOI
- https://doi.org/10.61325/rnzrw588
- Primary Topic
- Advanced Database Systems and Queries
- Type
- article
- Field-Weighted Citation Impact
- 0.00