Handling Session Classes for Predicting ASP.NET Performance Metrics

1
Handling Session Classes for Predicting ASP.NET
Performance Metrics
Ágnes Bogárdi-Mészöly, Tihamér Levendovszky,
Hassan Charaf Budapest University of Technology
and Economics, HungaryDepartment of Automation
and Applied Informatics
2
Outline

• Introduction
• Backgrounds
• Queueing model of web applications
• Model demonstration and validation in ASP.NET
  environment
• Model parameters estimation based on one
  measurement
• Model evaluation with the help of MATLAB
• Model validation
• Conclusions and future work

3
Introduction

• Web-based information systems
• An important factor performance
• Low response time
• High availability
• Certain level of throughput
• Properly designed performance model
• Performance prediction
• At early stages of the development process
• Prominent technology Microsoft .NET

4
Queueing Model of Web Applications 1

• Application network of M queues
• Application tier
• queue with processor sharing discipline
• A request can take multiple visits to each queue
• transitions to its successor and its
  predecessor
• Session-based workload
• infinite server queueing system
• Active session occupying one server

5
Queueing Model of Web Applications 2
6
Evaluation algorithm

• Mean-Value Analysis (MVA)
• Recursive algorithm
• Introduce customers one by one
• Terminate when all customers have been entered
• Visit numbers instead of transition
  probabilities
• Utilization law
• Utilization of the queues

7
Multiple session classes

• Model enhancement
• Incoming sessions classified into multiple
  classes
• Service times, visit ratios, user think time
• on a per-class basis
• Time and space complexities of MVA
• number of feasible population
• Approximate MVA algorithm in a few steps
• Balanced job bounds in one step

8
Model demonstration

• Three-tier ASP.NET test web application

9
Model parameters estimation and model evaluation

• Each page and class belonging to the
  presentation, business logic or database were
  measured separately
• Based on one measurement

Number of customers
MVA Approximate MVA Balanced job bounds
Max. number of customers
Response time
Number of tiers
Throughput
Average user think time
Tier utilization
Visit numbers
Average service times
10
Experimental configuration 1

• Application Center Test
• Warm-up time for the load to reach a
  steady-state
• Sleep times to simulate realistic usage of the
  application
• When we handle one session class, the number of
  simultaneous browser connections varied.
• When we handle multiple session classes, there
  were two classes a database reader and a
  database writer. The number of session of one
  class was constant 10, while the number of
  simultaneous browser connections of the other
  class was varied.
• Average response time and throughput were
  measured.

11
Experimental configuration 2

• Observed response times and throughputs

12
Model validation one session class 1

• Performance prediction with MVA algorithm
• Model predicts the response time and the
  throughput acceptably

13
Model validation - one session class 2

• Utilization
• presentation tier is the first that becomes
  congested
• database queue is the second (29)
• business logic queue is the last one (17)

14
Model validation - one session class 3

• Performance prediction with balanced job bounds
• Model predicts the response time and the
  throughput acceptably

15
Model validation - one session class 4

• Utilization
• presentation tier is the first that becomes
  congested
• database queue is the second
• business logic queue is the last one

16
Model validation two session classes 1

• Performance prediction with approximate MVA
  algorithm
• Model predicts the response time and the
  throughput acceptably

17
Model validation two session classes 2

• Utilization
• presentation tier is the first that becomes
  congested
• database queue is the second (76)
• business logic queue is the last one (15)

18
Model validation two session classes 3

• Performance prediction with balanced job bounds
• Model predicts the response time and the
  throughput acceptably

19
Model validation two session classes 4

• Utilization
• presentation tier is the first that becomes
  congested
• database queue is the second
• business logic queue is the last one

20
Conclusions and future work 1

• Model parameters have been estimated
• from one measurement
• MVA, approximate MVA evaluation algorithm and
  calculation of balanced job bounds have been
  implemented with the help of MATLAB
• Measurement process was executed
• in order to experimentally validate the models

21
Conclusions and future work 2

• Models predicts acceptably
• response time
• throughput
• tier utilization
• Model must be enhanced to handle
• limits of the four thread types
• in .NET thread pool
• global and application queue limit
• Model may be enhanced by other features

22

• Thank you for your attention.