Sometime ago we caught up with Dr. Leonid Grinshpan author of “Solving Enterprise Applications Performance Puzzles: Queuing Models to the Rescue”. Dr. Leonid Grinshpan (LinkedIn) is currently Technical Director at Oracle with a focus on Enterprise applications capacity planning, load testing, modelling, performance analysis, and tuning. Dr. Leonid has a few decades of experience on two complementing areas: computer science and information technology (IT) engineering.
Dr. Grinshpan also holds a Ph.D in computer science. He has worked on over 200 capacity planning and performance tuning projects for Fortune 500 customers over the past eleven years. Leonid has had an illustrious and impressive career. He has been the recipient of the highest award in the USSR for excellence in IT engineering, “Award of the Counsel of Ministers of the USSR” which was awarded to him for design and implementation of an open CAD system for microprocessor based products. Over the years Dr. Grinshpan has tested out his entrepreneurial skills by co-founding the first Belarusian-American Joint Venture Software Security Belarus which was acquired in 1997 by USA company Rainbow Technologies.
We asked Dr. Grinshpan a bunch of questions to get to know him better and understand the thought process behind his book, “Solving Enterprise Applications Performance Puzzles: Queuing Models to the Rescue”.
Question – Tell us more about yourself, what you do for Oracle and what is a day in the life of “Leonid Grinshpan” at Oracle
I have been working on capacity planning and performance tuning of Oracle enterprise applications for more than 16 years. My professional career has been characterized by constant traveling with great amount of face to face interactions with Oracle customers around the world. The latter gives me a tremendous opportunity to work on identification and solving of countless performance issues on enterprise applications of different types. The former has also been a source of life’s pleasantries such as as airline miles and hotel points.
Question – You have an impressive CV (LinkedIn – http://www.linkedin.com/in/leonidgrinshpan) with over 200+ projects that you’ve participated in or lead over the years. Could you share with our readers highlights of a few challenging projects that come to mind and what from your perspective made the difference in being able to deliver the outcomes.
If there are the difficulties in the projects, they originate not from the peculiarities of the production deployments and user’s workloads, but from use of incorrect performance troubleshooting methodology employed by the various support or performance teams. As such, to deliver a system that meet a Service Level Agreement, I have to succeed in working with people. I have to propose the right way of actions and to engage my colleagues to follow through.
Question – Given the decades of experience you have, what according to you are the most frequently made mistakes from a SPE (Systems Performance Engineering) standpoint that you have seen customers make time and again.
As I just pointed out, it is wrong performance troubleshooting methodology.
Question – You have recently published a book on Performance Modelling titled, “Solving Enterprise Applications Performance Puzzles: Queuing Models to the Rescue”. What made you write the book.
I strongly believe that SPE is not an art form, but an engineering science. SPE as a science has formal foundations; in my book I discuss one of them which help to understand performance malfunctions as well as to predict when they might happen.
Question – What kind of reader is the, book “Solving Enterprise Applications Performance Puzzles: Queuing Models to the Rescue” targeted at.
I want to reach out to performance engineers working on sizing, deployment, tuning, and maintaining enterprise applications. The book’s audience also includes system analysts, system architects, as well as developers adapting on deployment stage taken “of the shelf” enterprise applications to corporate business logistics and historical data. I hope it can be of value to system users in defining application workload. Project managers and sales professionals will find useful system sizing methodology.
Question – What are the various Performance Modelling techniques you have played around with over the years and which ones have you found to be most useful.
I studied different performance analysis approaches (many of them are described in a book “Performance Engineering of Software Systems” by Connie U. Smith). Taking into account practical limitations of my projects, I make use of queuing network based analytical modeling techniques for modeling of enterprise applications. The real-world limitations I have to deal with are: tightly packaged and short project time and limited precision of input data for modeling.
Question – Performance Modelling you will agree is a black art. It’s much more than just pure science and also requires a lot of experience including gut feel to get right. How realistic is it for a reader to pick up your book and get started with the art of Performance Modelling
This is where I disagree with you. Art is an expression of human creativity and imagination, a product of unrestricted flight of a fantasy. Contrary, science is a structured body of knowledge “that can be rationally explained and reliably applied “ (http://en.wikipedia.org/wiki/Science). We, performance practitioners, have to strive to promote and further develop SPE as a science – that is the only way to ensure that SPE will be capable to solve complex and practically important tasks. In my book, using system models, I illustrate root causes of performance deficiencies and how they can be cured. Modeling thinking but not a gut feeling will drive SPE.
Will reading of my book jump start a reader’s foray into system modeling? It definitely depends on a reader’s background and practical needs. I hope my book efficiently promotes modeling thinking as one of the foundations of SPE and paves the way to scientifically based methodology of system capacity planning and tuning.
Question – Is it practically feasible to model application performance using analytical modelling techniques and at what point during the development cycle are analytical modelling techniques most useful as an aid to understand/validate system performance.
My modeling experience presented in the book includes details of numerous successfully executed enterprise application capacity planning projects. Queuing networks are representative abstractions of enterprise applications, and analytical model solving is a right mechanism to use in corporate environment that puts strict limits on project time and data availability.
Question – What can our readers expect to take away from reading, “Solving Enterprise Applications Performance Puzzles: Queuing Models to the Rescue”.
They will be able to abstract from the complexities of a real system and create its representative models which will help to expose and eliminate the causes of capacity limitations and performance issues.
Question – In addition to Performance Modelling using Queuing Theory Modelling what are the other Analytical modelling techniques you would recommend that Performance Engineers should consider.
The biggest challenge in implementation of enterprise application modeling is a quantitative characterization of application transactional workload that represents input data for the models. I would recommend perusing my article http://tinyurl.com/k4fv95e that provides a road map to workload characterization and its prediction by:
- ‐ Identifying the constituents of EA transactional workload and specifying the metrics to quantify it.
- ‐ Reviewing the technologies generating raw transactional data.
- ‐ Examining Big Data Analytics ability to extract workload characterization from raw transactional data.
- ‐ Assessing the methods that discover the workload variability patterns.
Question – What are the other books you would recommend to our readers who have interests in learning more about details of Queuing Theory and other analytical modelling techniques.
A comprehensive assortment of the links can be found here http://tinyurl.com/oygw7oq
Question – Are you considering writing any other books focused on Statistical & Simulation modelling techniques for purposes of Performance Analysis.
I’m researching a topic “Enterprise Application in the Cloud – Capacity planning and Performance Analysis”. I’m publishing the articles and delivering presentations along the way. I do not know if the journey will result in a book, time will tell. My methodology is based on analytical approach to modeling for the same practical reasons I stated answering your previous questions.
Question – What message do you have for our readers at Practical Performance Analyst
Approach SPE as an engineering science; you will acquire and become proficient in right performance troubleshooting methodology. That will ensure your success in solving performance puzzles.
Trevor W – Thanks for taking the time and sharing with us your thoughts and experience on Systems Performance Engineering. We are glad to have you part of our family here at Practical Performance Analyst including the privilege to collaborate with you. Thanks again for all your support and best wishes.