Persisting Weakly Relational Data in NoSQL Stores

OData support
Dr. Ekler Péter
Department of Automation and Applied Informatics

Maintaining consistent state in distributed systems poses great problems for system architects. Cloud computing platforms, like OpenStack, provide REST APIs (Representational State Transfer Application Programming Interface) to query the state of the hosted/provided resources. Cloud management systems use this information to determine current state, and also to make decisions about operations. For effective functioning, this data should be available and up-to-date at all times. A trivial solution would be to get fresh information from the cloud on demand, but it is impractical and not possible for several reasons: it can take a long time to get response for queries that would slow down the executed orchestration processes significantly, and frequent queries can also be mistaken for an attack by the cloud platform, and in response it may prohibit further access. Therefore, a cloud management system may build and use a cache as a practical solution for this problem, which should be kept up to date with a consistent query strategy. In this model, the resources are represented by entities, and the connections between those entities represented as relations. The expiration of entities is vital in this concept. If a resource is not refreshed for a certain amount of time, it has probably been removed from the system, thus, no longer available. The detection of such circumstances is not always trivial, which was the reason to create the definition of Weakly Relational data. In this thesis the definition of Weakly Relational data is given, and examples are also provided. After selecting an appropriate database for storing Weakly Relational data and investigating the cloud environment, an architecture and implementation is provided for the concept. The capabilities of the developed component are demonstrated with an example, and measurements are also presented.


Please sign in to download the files of this thesis.