Finding bottlenecks in IT systems

Abstract

Nowadays, more and more segments of the information technology (such as the networks, the Internet, the World Wide Web) become essential in the everyday life. We get new IT- systems that are often robust and comprehensive. Companies need to use information systems to keep abreast of the times. In the past information systems have appeared stand- alone, but now they often use networking solutions for office work, document management etc. Although the reliability and availability are very important aspects of the services the efficient provision of these attributes with the smallest hardware, software and money investment is also crucial. IT systems are dimensioned in that manner that their load is light or medium in most of their life, however, they suffer heavy load in a small fraction of their operation. Since this heavy load can cause critical situations our main goal is to explore and identify the bottlenecks of systems, and to give proposals how to avoid the problems. We use a test environment where we measure the load factors of each threads or processes. Our measurements cover memory usage, CPU usage and I/O data traffic. Based on the observed data and their operation environment, tendencies are identified that help system operators to prevent systems from overloaded situations.