- This is the code used for the MongoDB part in my second bachelor thesis.
- There is also a MySQL and a Couchbase part.
- The thesis itself is written in German. You can find the abstract below.
To satisfy the needs of massive scalability regarding to parallel read and write queries new database systems have evolved. This paper introduces document stores and compares them with traditional relational databases. It outlines the pros and cons of both system types and gives a comparison on the most common data models, which are the embedded and referenced document model for the document stores and normalized and denormalized table model for the relational databases. This paper also discusses transactions and how to implement them in the application layer in case the database does not have a support for transactions. A test environment is set up and a sample application is used in order to measure the response time of the systems. MongoDB and Couchbase are used as representatives for document stores and MySQL in combination with InnoDB, MyISAM and NDB as storage engines is used as a representative for relational databases. The sample application mimics a heavily used blogging platform with users writing and reading blog entries and comments and making use of a “Like” button. The results show that the embedded document data model as well as the denormalized table data model are much more responsive to read queries then the referenced document data model and the normalized table model, but with a serious performance bottleneck on write queries. The normalized table data model shows the worst performance. Furthermore the denormalized table model has to store the same information multiple times, which can lead to inconsistent data. Because the blogging platform does not have the requirement to use transactions, the best database solution for such a big data application is the document store with a combination of the embedded and referenced data model. Therefore the paper confirms the positive impact of document stores to the current database market.