Estudo e aplicação de tuning em sistema gerenciador de banco de dados orientado a grafos

Abstract

The volume of data stored in database grows exponentially every day. For the management of large amounts of data, the relational model has limitations related to scalability, which led to the emergence of new solutions, as is the case of NoSQL Database Management Systems. When there is a need to manage highly connected data, such as social networks, the most suitable storage model to use is the graphoriented one. However, even though it performs satisfactorily in his application area, there is the possibility of improvements in terms of performance, thus bringing competitive advantages to companies. One way to improve the performance of Database Management Systems is adjust and modify the management system, the operating system, base schema, or available hardware. The objective of this work is to apply tuning in a graphoriented Database Management System. For this, a survey of possible techniques to be applied was carried out together with the creation of a test environment. The experiments demonstrate a reduction of more than 80 minutes in the execution time of 57.14% of the workload queries, when changing the RAM memory usage settings. When it comes to indexes, the addition of new types reduced the execution time of 66.87% of queries in the workload used by around 72 minutes.