Predicting query response time supports capacity planning and the prioritization of database design changes, yet many published prediction models rely on benchmark workloads and opaque learned models. This study reports an empirical analysis of a production PostgreSQL transaction system in a glass manufacturing company, using multivariate linear regression as a transparent baseline. Response time (s) was modelled from four design-level variables: number of tables, number of indexes, database size (MB) and query complexity (number of joins), using 45 observations collected during the 2016–2025 period via direct measurement from production logs. The fitted model reported R² = 0.92 and RMSE = 0.12 s on chronological hold-out data. The four predictors were strongly intercorrelated (r = 0.92–0.99) and each correlated 0.97–0.98 with response time, so individual coefficients cannot be read as separate causal effects. The results are associations observed in one system and have not been validated on other workloads or configurations. The model is proposed as a transparent baseline for workload planning, and optimization recommendations are presented as hypotheses that require before-and-after testing.
Arahkan kursor atau ketuk sebuah nama untuk menyorot rekan penulisnya. Klik dua kali untuk membuka halaman penulis.
Soegijapranata Catholic University
Organisasi berbeda
Organisasi sama
Afiliasi tidak diketahui
Garis lebih tebal = lebih banyak artikel bersama · lingkaran lebih besar = lebih banyak artikel
Tampilkan sebagai tabel
Rekan penulis
Organisasi
Artikel bersama
Afiliasi berasal dari OpenAlex; metadata OAI-PMH jurnal tidak memuatnya. Penulis tanpa ID OpenAlex dicocokkan berdasarkan nama, sehingga dua orang bisa tergabung atau satu orang bisa terpisah.