Polypheny-KV: A Transactional Key-Value Store for Polypheny-DB (Master Thesis, Finished)


Christian Frei


Today's requirements on Big Data analysis are very high because this data comes in many shapes and forms. To meet this challenges, a new form of Database Management Systems has been developed, so called Polystores. Polystores combine different kinds of databases, from Relational Database Management Systems and NoSQL systems up to column stores.

In this Master's Thesis we introduce Polypheny-KV, a transactional and relational in-memory Key-Value Store. It combines the speed of a traditional Key-Value Store with the advantages of a Relational Database Management System.

Polypheny-KV will be implemented as new data store in Polypheny-DB, which is a recently developed Polystore. We will evaluate the system with TPC-C and the Gavel benchmark. Our goal is to outperform these benchmarks compared to the previous benchmarks of Polypheny-DB because our new data store Polypheny-KV mainly addresses OLTP queries, which are currently the only bottleneck of Polypheny-DB.

Start / End Dates

2017/11/15 - 2018/05/14


Research Topics