At this page you’ll find content made available by Dr. Jayanta Choudhury on topics related to SPE (Systems Performance Engineering. Dr. Jayanta Choudhury is a technology researcher at TeamQuest Corporation, focusing on capacity planning and performance modeling for IT resource optimization. All the articles listed on this page were written by Dr. Jayanta Choudhury for Practical Performance Analyst while the presentations were created for various talks that Dr. Jayanta Choudhury has delivered at conferences, workshops and seminars across the USA.

We hope you find this section interesting and helpful in your quest to learn more about SPE (Systems Performance Engineering) across the SDLC (Software Development Life Cycle). All the content from Bob’s book has been published with permission. Listed below are the various resources that Dr. Jayanta Choudhury has put together over the years:

1) Presentation – Back to the Future of IT Resource Performance Modelling & Capacity Planning

This material was presented at the 5th International Conference of Education, Training and Informatics (ICETI 2014) held in Orlando, Florida, March 4-7 2014. The glaring failure of `healthcare.gov‘ is directly related to the area of performance modeling and capacity planning of computing systems. The fundamental principles of modeling of computing systems using Queuing Network Theory are based on the assumption of steady-state flow condition of the computing systems after an initial unsteady system-start state. The advent of virtualized systems with automated movements of virtual machines between physical machines and cloud computing paradigm are making computing systems so dynamic that a long enough duration of steady-state to collect data of sufficient statistical accuracy to feed to Queuing Network model of computing systems are less likely.

Please click here to download the presentation.

2) Presentation – Theoretical Model and Data : Truth ?

In this presentation Dr. Jayanta Choudhury compares various theoretical models for determining the performance and scalability of computer systems. This material was presented at the St. Louis CMG meeting held on April 8th 2014. The crux of the presentation is a comparison of the various scalability modelling techniques which include:

  • Amdahl’s non linear growth model of parallel computing
  • Gustafson’s law
  • Super serial scalability law
  • Sun and Ni’s law
  • Universal Scalability Law by Dr. Neil Gunther

Various scalability models are available today like Amdahl’s Law, Gustafson’s Law, Super-serial Scalability Law, Sun and Ni’s Law and Universal Scalability Law etc. Experimental data does not always support all of them. Deeper understanding of the abstraction of various factors, influencing performance of computing systems, are presented to understand the situation if incompatibility between data and one of the models are encountered. It is shown how the models can be effective tool if used appropriately by choosing proper computing environment that complies with the assumption of the models.

Please click here to download the presentation.

3) Presentation – SSL v/s USL : Using Least Square Error Principle and the Machine Repairman Model

This material was presented at the CMG 13 Performance & Capacity Conference held at La Jolla, California from Nov 5-7.  In this presentation Dr. Jayanta Choudhury compares the SSL (Super Scalability Model) against Dr. Gunther’s USL (Universal Scalability Model). Dr. Choudhury also focuses on analyzing the weaknesses present in both the models with regards to predicting system scalability. Finall Dr. Choudhury moves onto analysis of the “Machine Repairman model” including the “Least Square Error Principle”.

Please click here to download the presentation.

4) Presentation –  A More Robust Regression Methodology to Estimate the Parameters of Super-serial Scalability Law from Noisy Data

A methodology to estimate the parameters of Super-serial Scalability Law (SSL) was first proposed in 2001 by Dr. Neil Gunther. SSL was substituted by Universal Scalability Law (USL) later but it is shown here that SSL displays better performance even when USL fails. Some theoretical errors of the first methodology along with a new methodology, circumventing those theoretical errors, were reported recently by the author. More improvements to the methodology of the author is reported here to get better results from noisy input. The proposed improvements are applied to sample noisy performance data.

Please click here  to download the presentation.

5) Presentation – Mythbuster for the Guerrillas

A methodology to estimate the constant parameters of Universal Scalability Law (USL) has been proposed by Dr. Neil Gunther. Certain theoretical inadequacies of the subject methodology are found when applied to some typical sample data sets. An improved methodology is proposed to circumvent the problems. USL is also compared against Super-serial Scalability Law (SSL) for accuracy. SSL is found to be more robust than USL in case of noisy or insufficient input data.

Please click here  to download the presentation.

6) Paper  – Re-evaluating “Evaluating Scalability Parameters: A Fitting End

The article, “Evaluating Scalability Parameters:A Fitting End” by Dr. Neil J. Gunther, contains  a step by step method to estimate the parameters, σ and λ, of the Super-serial Scalability Law (SSL) using Microsoft Excel. Further investigation of the scheme shows some deviations from expected behavior. In this white paper, the root cause of the deviations from the expected results
are explained and an improved scheme is proposed for getting more accurate estimates.

Please click here  to download the presentation.


Dr. Jayanta Choudhury (LinkedIn), is a technology researcher at TeamQuest Corporation, focusing on capacity planning and performance modeling for IT resource optimization. He finished graduate school with a PhD degree from the University of Louisiana at Lafayette, in 2008. He has been working in the area of capacity planning and performance modeling of computing systems since 2007. His research interests include performance modeling, capacity planning, operations research, high performance computing, algorithm development, data analysis, numerical analysis, and numerical solution of PDEs, ODEs and their applications. He focuses to understand the fundamental principles behind the algorithms and methodologies used in the area of capacity planning and performance modeling. You can reach Dr. Jayanta Choudhury at jayanta.choudhury@gmail.com.

Thank You For Visiting & See You Soon

Practical Performance Analyst - NewsLetter

Signup and Stay In Touch

Signup to the Mailing List here at Practical Performance Analyst and Stay In Touch. We promise not to SPAM you.