Scalability analysis of multi-tier distributed software patterns using layered queuing networks

Factors such as multiple jobs, multiple heterogeneous resources, variety of communication mechanisms and QoS issues make scalability analysis of distributed systems more complex than that for parallel systems. In this paper, we analyze scalability of distributed multi-tier software patterns such as towers and pyramids. The paper gives an insight into the scalability behavior of these patterns with the help of the extensive sets of experiments that we have conducted to study the effect of different scaling strategies and their respective parameters on the scalability of a particular pattern. Our results show that the choice of a scaling strategy should not only take into account the pattern being scaled up but also factors such as the scale factor to which the scalability is desired and the size of the pattern. For pyramid pattern it has been observed that often it is desirable to do simultaneous scale up of multiple layers as opposed to scaling up one layer at each scale factor. A suitable scaling strategy has been proposed for pyramid pattern that gives a cost effective increase in performance for a wide range of scale factor