Large scale application design and development involve some critical decisions, and one of the most important issues that a ect software application design and development is the technology stack used to develop an extensive system. In a JPA API, response time is often a measure of how quickly an interactive system responds to user input. Persisting framework, such as Object Relational Mapping (ORM) are applied to manage communications between an object model and data model components and are vital for such systems. Hibernate is considered the most e ective ORM framework due to its advanced features, and it is the de-facto standard for Java Persistence API (JPA)-based data persistent frameworks. This thesis comprises a review of the most widely used JPA providers, particularly frameworks that provide JPA support such as Hibernate JPA, EclipseLink, OpenJPA and DataNucleus JPA. In current java programming, APIs based on persistence and performance are integral aspects of an application. Performance analysis of the above four JPA implementations is based on the ORM framework that contributed most signi cantly to discovering the challenges and veri ed the programming considerations in the language. For a large-scale enterprise, working on JPA is always tedious due to the potential pressures and overloads of the implementations, as well as the comprehensive guarantee, that is required while adopting the technology. A JPA implementation continually persists data into the database at runtime, by managing persistence processes through interfaces and classes, that often needs optimization, to provide performance-oriented results at heavy loads. Therefore, in this thesis a detail feature analysis was performed, before the performance analysis. To enhance the comparison of the persistence framework, an extended experiment with a cloud database using Database-as-a-service (DBaaS) versus Physical Persistence was performed, using a comparative approach for all four JPA implementations. Di erent SQL queries on cloud versus physical persistence for JPA applications were measured using CPU, GC, and threads (live and daemon). Finally, a statistical analysis was performed using the Pearson's correlation coe cient and a steady/start-up phase.