SlideShare uses cookies to improve functionality and performance, and to provide you with relevant advertising. If you continue browsing the site, you agree to the use of cookies on this website. See our User Agreement and Privacy Policy.
SlideShare uses cookies to improve functionality and performance, and to provide you with relevant advertising. If you continue browsing the site, you agree to the use of cookies on this website. See our Privacy Policy and User Agreement for details.
Successfully reported this slideshow.
Activate your 30 day free trial to unlock unlimited reading.
1.
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
Adaptive Learner
Arun Kumar Dhakshinamoorthy
2.
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
AGENDA
What is Adaptive Learner
Adaptive Learner in Performance Testing Lifecycle
How it works ?
Key Benefits & Conclusion
Anomaly Detection
Backpropagation using python
Determination of Machine Learning Model
3.
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
What is Adaptive Learner
Adaptive Learner is a
automated Solution
which provides
constant feedback in
performance testing
Automatically
redefining the testing
strategy
Goal is to find a
Performance Problem
where application
behave different for
various combination of
data
Identify Bottlenecks or
Hot spots where
Performance is limited
due to Software or
Hardware
configurations
Sample
Implementation
Machine Learning
process is used for
detecting Memory
based Anomalies to
configure and tune
JVM Parameters
4.
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
Adaptive Learner In Performance Testing Lifecycle
Identity NFR’s
Design Work
Load Mix
Design Test Script
and Scenario
Run Test Using
Load Testing Tools
Analyse Test
Results
Design Build Testing Test Analysis
Performance Tuning
JVM using ML
based Algorithm
Re design Workload
Mix /data
Adaptive Learner
Updated JVM
Configuration
5.
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
How It Works ?
• Different test data sets are identified based full volume of data and predefined Scenario are created with the
datasets for Load testing
Identifying Data and Scenarios
• Anomaly are identified in memory pattern and JVM Parameter which needs to be optimized are identified and tuned
• Final testing is performed on correct dataset exercising the application methods effectively and running test against
optimised JVM parameters
Tuning JVM Parameters
• The Adaptive Learner program analyse the datasets used and capture the relevant metrices like CPU, Heap Memory
patterns from the applications server using APM tools.
• Anomaly in memory usage patterns are analysed using machine learning based algorithm in the learner.
• High utilization methods in the application program are identifies based on the usage of different datasets and based on
which the rules are created as a feedback and test scenario is modifies for better performance load.
Analysis using Adaptive Learner
• Simulate Load testing with the subset of data using industry standard load testing tools like LoadRunner
Performance Test Execution
6.
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
Anomaly Detection
An anomaly is referred to as any abnormal behaviour in application performance
Various reasons for anomaly
• Transaction Response time going beyond business approved SLA
• High Utilization of CPU
• Memory going beyond 75% of Threshold
• Frequent occurrence of major Garbage collection
The learner will concentrate on abnormality induced due to violation in memory consumption
• Historical data samples are collated and stored as training data for
ML Framework
• The Framework is modelled to achieve higher success probability
with the help of extensive training dataset
• Output from the model serves as testing data for ML framework
whenever an abnormality is detected in the application behaviour.
• GC times are collected as test data using automated utility and fed
into source database for ML Model.
Data Collection for Machine Learning Framework
7.
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
Determination of the Machine Learning Model
Most suitable approach to achieve the relationship between the JVM metrics and the garbage collection time is to implement an
Artificial Neural Network (ANN) using backpropagation algorithm
An Artificial Neural Network is a mathematical structure which is widely used to determine a relationship between the input and
output parameters.
The aim is to find the optimal value of the weightage of the neural network to get the desired output
What are the Steps involved in designing a back propagation algorithm ?
• Defining Input Layer
• Defining Hidden Layer
• Defining Output Layer
• Assigning Weight
• Neuron Activation.
• Neuron Transfer
(Sigmoid Activation
Function)
• Forward Propagation
• Transfer Derivative.
• Error Backpropagation
• Update Weights.
• Train Network using
Gradient decent
Forward PropagateInitialize Network Back Propagate Error Train Network & Predict
8.
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
Backpropagation using python
The input layer for the ANN will consist of the most important
JVM metrics and Load applied such as
• Min Heap Size (m1)
• Max Heap Size (m2)
• New Size(m3)
• New Ratio(m4)
• Min Heap Free Ratio(m5)
• Max Heap Free Ratio (m6)
• Max New Size (m7)
• Throughput(M8)
• Output parameter is the predicted garbage collection time.
• ML Framework will determine a relationship between the
input and output layer by applying the training data to the
ANN model.
• The relationship can be represented as f(GCT) = f(m1, m2, .... ,
mn).
9.
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
Key Benefits
Application is Simulated with
Different subset of data and load
Identifying bottle neck in the
system for specific data
conditions
Automated Feedback of system and MI based JVM Optimization
improves application performance
Tuning based on the Feedback
helps in Optimising the system
performance
10.
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
Conclusion
• Identify
bottleneck
• Identify methods
which are heavily
used base on
different data sets
• Automatic
Feedback
• 50% reduction in
manual work
• Identify Issue faster
• Run different
scenarios
automatically
• Optimise performance by
tuning JVM parameters
• Automated analysis of data
shared by APM tools using
ANN Algorithm
11.
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
#ATAGTR2020
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial
channels(Provided due credit is given to me/us)
0 likes
Be the first to like this
Views
Total views
86
On SlideShare
0
From Embeds
0
Number of Embeds
0
You have now unlocked unlimited access to 20M+ documents!
Unlimited Reading
Learn faster and smarter from top experts
Unlimited Downloading
Download to take your learnings offline and on the go
You also get free access to Scribd!
Instant access to millions of ebooks, audiobooks, magazines, podcasts and more.
Read and listen offline with any device.
Free access to premium services like Tuneln, Mubi and more.