Comparative Analysis of the Performance of Prisma ORM and the Native pg Driver When Working with PostgreSQL in Node.js Applications

Abstract

The choice of database access tool is one of the fundamental architectural decisions in Node.js server-side application development. ORM tools, including Prisma, simplify interaction with relational databases through a type-safe interface, yet introduce an additional abstraction layer whose impact on query performance remains insufficiently studied in the context of current tool versions. In 2025, Prisma underwent an architectural transition from an internal Rust engine to a native TypeScript engine, which fundamentally changed the performance profile of the tool and rendered the results of previous comparative studies invalid for extrapolation to the current state of the ecosystem. This study presents a controlled benchmark of Prisma ORM version 7 and the native pg driver version 8 operating against PostgreSQL 17 in a Node.js environment. Measurements were performed in isolation from the HTTP layer as a standalone Node.js process outside the context of any web framework, on a real TypeScript application database schema with 10,000 records. For each of four operation types – simple SELECT, filtered SELECT with pagination, INSERT, and a JOIN query across four tables – 1,000 iterations were executed with a 50-query warm-up. Mean, median, minimum, maximum, and standard deviation of latency were recorded. The results show that the pg driver demonstrates lower mean latency across all four scenarios – ranging from ~25% for INSERT to ~47% for the JOIN query. The observed gap is substantially smaller than reported in previous studies, confirming a qualitative shift in Prisma's performance following the transition to the TypeScript engine. The absolute difference between the approaches is 0.5–1.1 ms, which is statistically significant but practically negligible for the vast majority of production applications. Result stability was confirmed by a repeated run on a dataset of 100 records. The findings provide a practical basis for informed selection of a database access tool when developing TypeScript applications on Node.js with PostgreSQL.