An Empirical Evaluation of Object-Relational Mapping Frameworks on the Performance of .NET Applications

Abstract

Object-Relational Mapping (ORM) is a widely adopted programming technique that enables software engineers to perform database operations using object-oriented programming (OOP) languages instead of SQL statements. By providing a mapping layer between object models and relational databases, ORM frameworks reduce development time and shield developers from database complexity. In the. NET ecosystem, common ORM frameworks include Entity Framework Core and NHibernate. Despite their popularity, the performance implications of these frameworks compared to native SQL clients have not been thoroughly investigated. This research presents a comprehensive performance evaluation of Entity Framework Core, NHibernate, and the Native SQL client (ADO. NET) across Microsoft SQL Server (MSSQL) and MySQL databases. A. NET Core application was developed to execute various CRUD and complex query operations on both databases. Results reveal that MySQL operations generally require more execution time than MSSQL. Interestingly, on MySQL, ORM frameworks performed comparably to, and in some cases better than, the Native SQL client. Conversely, on MSSQL, ORMs exhibited significant overhead, with complex queries taking more than five times longer than their native counterparts. These findings provide practical guidance for software engineers and architects in selecting appropriate data access strategies based on database choice and application requirements.