Abstract or Summary:
In this paper, we argue that a combination of various monitoring techniques is essential for identifying performance bottlenecks in distributed applications that execute on locally networked computer systems. We propose an integrated tool that is comprised of an application monitor, an operating system monitor, and a network and host monitor. Each of these monitoring techniques is well established on its own. However, we show that it is the integration of these techniques that is often crucial to finding the real causes of performance problems. We understand integration as the seamless merging, synchronisation, and presentation of performance data from all monitoring sources. The proposed tool allows both automatic and interactively controlled measurements, whereby automatic measurement mode is intended as a first monitoring step in order to discover possible performance bottlenecks with minimal effort.

PasswordResearch.com Note: One of the distributed software case studies in this paper is password cracking software. The authors identify and measure the impacts of different cracking process optimizations.