• Christian Klim Hansen
  • Nedim Cokljat
4. semester, Software, Kandidat (Kandidatuddannelse)
Research into heterogeneous database systems and alternative database models has become more and more relevant as the amount of stored data continues to increase. Specialized database models are created to perform efficiently when performing a certain workload. Graph databases are well suited for traversal queries, such as nding the shortest path. The typical information queried for in graph databases is meta data, such as finding how nodes are related to each other trying to find a pattern in the graph structure. Relational databases are more powerful when it comes to querying for data about specific entities, since over the years a lot research has been put into optimizing indexing in relational databases. But when it comes to finding associations between entities joining of tables is required, and joining complex associations can be slow, and that is where graph databases outperform relational databases since they do not have to join.
Bridge-DB is a fault tolerant middleware system that connects multiple different database models and enables users to read and write to externally connected heterogeneous databases while ensuring eventual consistency and availability. Bridge-DB is tested with an external Neo4J and PostgreSQL database using full redundancy, but is designed to be a general framework for working with multiple heterogeneous databases. The tests illustrate the need for Bridge-DB by showing how different query types are executed faster and slower depending on the selected database model. Our experiments further document, that Bridge-DB upholds the BASE properties while supporting all create, read, update and delete operations.
SprogEngelsk
Udgivelsesdato2015
Antal sider19
ID: 213631644