Abstract
NoSQL data stores are becoming increasingly popular in application development. These systems are attractive for developers due to their ability to handle large volumes of data, as well as data with a high degree of structural variety. Typically, NoSQL data stores are accessed programmatically. Due to the imminent lack of standardized query languages, building applications against the native ...
Abstract
NoSQL data stores are becoming increasingly popular in application development. These systems are attractive for developers due to their ability to handle large volumes of data, as well as data with a high degree of structural variety. Typically, NoSQL data stores are accessed programmatically. Due to the imminent lack of standardized query languages, building applications against the native interfaces of NoSQL data stores creates an unfortunate technical lock-in. To re-gain platform independence, developers turn to object mapper libraries as an additional level of abstraction when accessing NoSQL data stores.
The current market for Java object mappers that support NoSQL data stores is still volatile, with commercial and open source products competing for adoption. In this paper, we give an overview on the state-of the-art in Object-Relational Mappers that can handle also NoSQL data stores, as well as dedicated Object-NoSQL Mappers. We are able to show that choosing the right object mapper library is a strategic decision with far reaching consequences: Current mappers diverge in the NoSQL data stores that they support, in their features, their robustness, their truthfulness to the documentation and query standards, and ultimately, in the runtime overhead that they introduce. Especially in web development, runtime overhead is a crucial aspect contributing to the application latency, and ultimately, the user experience. By shedding light on the current market, we intend to provide software architects with the necessary information to make informed decisions.
More literature (via CORE)