Query rewriting for continuously evolving NoSQL databases

Abstract

In agile software development settings, applications are typically backed by schema-flexible NoSQL databases. New application code frequently implies data model changes to the effect of multiple schema versions within the NoSQL database. Here, a query rewriting approach can handle the issue of how to access legacy data, otherwise datasets in previous schema versions would seem to disappear for the application. Our NoSQL query rewriting approach for multi-versioned databases takes evolution operations into account, their reverse operations as well as the heterogeneity of data. For that purpose we specify four NoSQL heterogeneity classes from relational up to completely unstructured NoSQL records. Furthermore, we propose a NoSQL query rewriting algorithm that generates subqueries compatible to all existing structural versions.