storr
Conceptual Model
Storr is a NoSQL store which is intended to provide easy storage of arbitrary tuples. Unlike many NoSQL stores it is not based on a key-value storage model and instead supports the storage of arbitrary collections of tuples. As such it has features in common with NoSQL, relational and graph databases.
Features of the store include:
- logical structuring via repositories and buckets
- support for both typed or untyped collections of tuples (via buckets)
- support for transactions
- support for object references
- a REST API
- a Java API
- optional structural type equivalence
- ability to store types
- indexing over arbitrary fields of the tuples
- the ability to provide different storage regimes (under development)
These features are described below.
A Storr database is partitioned into three levels: the Store which contains repositories which in turn contain buckets. The tuples that are stored in a Storr database are stored within buckets.
This structure was created to permit structuring without the introduction of too much complexity. Storr was originally created to store Birth, Death and Marriage records in the Digitising Scotland project. As an example of the use of Storr, in our test system we had two sets of records - one from the Isle of Skye and another from Kilmarnock. To support such a setup we created two repositories called Skye and Kilmarnock respectively each of which would contain three buckets, each containing the Birth, Death and Marriage records. An initialised Storr store to support this structure would look like this:
