Decibel : transactional branched versioning for relational data systems

[摘要] As scientific endeavors and data analysis become increasingly collaborative, there is a need for data management systems that natively support the versioning or branching of datasets to enable concurrent analysis, cleaning, integration, manipulation, or curation of data across teams of individuals. Common practice for sharing and collaborating on datasets involves creating or storing multiple copies of the dataset, one for each stage of analysis, with no provenance information tracking the relationships between these datasets. This results not only in wasted storage, but also makes it challenging to track and integrate modifications made by different users to the same dataset. Transaction management (ACID) for such systems requires additional tools to efficiently handle concurrent changes and ensure transactional consistency of the version graph (concurrent versioned commits, branches, and merges as well as changes to records). Furthermore, a new conflict model is required to describe how versioned operations can interfere with each other while still remaining serializable. Decibel is a new relational storage system with built-in version control and transaction management designed to address these shortcomings. Decibel;;s natural versioning primitives can also be leveraged to implement versioned transactions. Thorough evaluation of three versioned storage engine designs that focus on efficient query processing with minimal storage overhead via the development of an exhaustive benchmark suggest that Decibel is vastly superior to and enables more cross version analysis and functionality than existing techniques and DVCS software like git. Read only and historical cross-version query transactions are non-blocking and proceed all in parallel with minimal overhead. The benchmark also supports analyzing performance of versioned databases with transactional support. It also enables rigorous testing and evaluation of future versioned storage engine designs.

[发布机构] Massachusetts Institute of Technology