| Abstract [eng] |
This thesis proposes a Metric Vault development methodology for Data Vault 2.0 data warehouses. Currently, Data Vault 2.0 data warehouses do not collect computer resource metrics, however, this information is very relevant when trying to solve data warehouse performance problems. This information could be used for data warehouse management to speed up the system, reduce resource usage, or reduce operational costs. To store such information Data Vault 2.0 architecture proposes a new component called Metric Vault, however, it does not provide guidelines on how it should be implemented. Thus, the main goal of this thesis is to facilitate the monitoring and analysis of computer resource metrics by proposing a Metric Vault development methodology. This thesis presents the Metric Vault development methodology, which details the whole Metric Vault development process from adding metric data sources to creating metric visualizations. The methodology has been developed by analyzing current Data Vault formation practices so that it could be integrated into current processes. A new Metric Vault development tool has also been created to automate the Metric Vault creation flow. The tool allows users to create their Metric Vault model, categorize the metrics, and generate DDL and ETL code, which creates a data warehouse structure and loads the metric data. An experiment has been conducted to evaluate the applicability of the proposed methodology and implemented tool. During the experiment a new Metric Vault was created to store the company’s server data, also a survey was published to collect the responses from Data Vault data warehouse specialists. The newly created Metric Vault stored about 13 million records and had 7 visualizations to illustrate different metrics. The majority of responses from the survey stated that the proposed methodology and the implemented tool can facilitate the creation of Metric Vaults for Data Vault data warehouses to help with metric monitoring tasks. |
