Analyzing and mining transactional data straight from the DBMS has become increasingly more popular, provoking research in read optimization for RDBMS. One branch of such research – column orientation claims significant improvement in... more
Analyzing and mining transactional data straight from the DBMS has become increasingly more popular, provoking research in read optimization for RDBMS. One branch of such research – column orientation claims significant improvement in read performance in comparison with row oriented DBMS. Having previously looked into the strengths and weaknesses of both approaches we take on studying the performance of commercial grade DBMSs, which employ the two views. The first step in this is examining their data models. We then develop a benchmark, which is subsequently used to measure each DBMS's performance. Evaluating the results we draw conclusions about each DBMS's suitability and main advantages over the other.
The focus of this paper is the Configuration Management Database (CMDB) as seen by Information Technology Infrastructure Library (ITIL). It aims at providing a conceptual and logical model for this database. The core theoretical CMDB... more
The focus of this paper is the Configuration Management Database (CMDB) as seen by Information Technology Infrastructure Library (ITIL). It aims at providing a conceptual and logical model for this database. The core theoretical CMDB concepts are considered and also many practical problems are discovered. Having uncovered the main obstacles and problems that an implementer faces, valid solutions are provided. As a result a logical model of a CMDB is proposed. Its practical value for implementation has been tested with real life data and satisfactory results are reached.
Most transactional database systems have been optimized for write performance. However, as analyzing and mining transactional data straight from the DBMS has become increasingly popular, read optimization has gathered significant interest... more
Most transactional database systems have been optimized for write performance. However, as analyzing and mining transactional data straight from the DBMS has become increasingly popular, read optimization has gathered significant interest recently. In this paper we compare the widely used row oriented DBMS with the column stores, which gain more and more momentum in the last several years. We focus on the data model of both systems as well as specific read optimization techniques. By analyzing previous performance studies we draw conclu-sions about the advantages and disadvantages of both systems in general and on analytical query loads.
Data mining has become more popular in recent years and with it the trend of analyzing transactional data straight from the DBMS. This trend is provoking the research into read-optimized database solutions. One branch of such research –... more
Data mining has become more popular in recent years and with it the trend of analyzing transactional data straight from the DBMS. This trend is provoking the research into read-optimized database solutions. One branch of such research – column orientation claims significant improvement in read performance in comparison with row ori-ented DBMS. Another branch is the growing number of NoSQL solu-tions. Having previously conducted research between the traditional relational DBMS and column-stores we take on studying the performance of NoSQL DBMSs in the face of MongoDB with the goal of comparing the three approaches. The first step in this is examining their data models. We then use our previously developed benchmark to measure each DBMS's performance. Evaluating the results we draw conclusions about each DBMS's suitability and main advantages over the other.