Performance Engineering

This open access book improves the users' skills needed to implement models for performance evaluation of digital infrastructures. Building a model is usually a relatively easy task, but making it an accurate representation of the phenomenon to be reproduced is a completely different matter. It is well-known that to increase the ability to build reliable models it is necessary to accumulate experience. The book addresses this need by presenting a collection of case studies of increasing complexity. Readers are introduced to the modeling process gradually, learning the basic concepts step-by-step as they go through the case studies. Queueing Networks are used to design the models solved with simulation and analytical techniques from the open source Java Modelling Tools (JMT). Among the models analyzed there are systems for optimizing performance, identifying bottlenecks, evaluating the impact of the variability of traffic and service demands, analyzing the effects of synchronization policies in parallel computing. Four case studies derived from real-life scenarios are also presented: a surveillance system, autoscaling load fluctuations, web app workflow simulation, and crowd computing platform. This book serves as a reference tool for graduate and senior-level computer science students in courses of performance evaluation and modeling, as well as for researchers and practitioners.

Explores the topic of modeling for performance evaluation of digital infrastructures Describes models which provide a good balance between accuracy of results, complexity & parameterization effort Serves as a reference tool for researchers interested in the performance evaluation of computer infrastructures This book is open access, which means that you have free and unlimited access

Autorentext

Giuseppe Serazzi is emeritus professor at Politecnico di Milano, Italy. Over his long career he has co-authored 150+ papers and 14 books. His research interests include performance engineering of enterprise infrastructures, modelling techniques and tools for capacity planning of very large systems. In the 90 s he was the initiator, and still is one of the coordinators, of the JMT open-source project (Java Modelling Tools for performance engineering of computing infrastructures). From 1993 to 2000 he served as editorial board member of the Elsevier journal "Performance Evaluation".

Inhalt

The process of Modeling.- Systems with homogeneous workloads.- Systems with heterogeneous workloads.- Impact of variability of interarrival and service times.- Parallel computing.- Reference models.