Performance Testing of Large Scale Distributed System

Traditionally performance testing approach usually follows Identification of Key Scenarios, Setting up the Load Environment, Designing the Scripts, Generating load, Monitoring and at last Analysis and Reporting.  It works for most of the system but it is completely a different ball game when conducting performance testing of large scale distributed system.  In this article, some of the key characteristics and challenges will be discussed for performance testing of large scale distributed system.

Key Characteristics

High Volume

»        Terabyte of transactional records in database

»        Network throughput in gigabits per second

High Transactions

»        Millions of transactions per second from end users

»        Millions of transactions in database due to few triggers (e.g. large report generation due to batch processing)

High Concurrency

»        Huge user base accessing simultaneously (e.g. Facebook)

Geographically Distributed

»        Traffic from all over the world

High availability

»        Huge revenue loss and complaining customers due to unavailability

Huge Data Analytics

»        Big data, data warehousing

 

Key Challenges

Proposed Solution

High cost for test environment setup

»        Production or staging environment

»        Scaled down environment

High cost for load generation environment

»        Cloud based load generation tool

High license cost for tools and utilities

»        Open source load generation / monitoring tools

»        Use pay per service if number of runs are less

Production like environment configuration

»        Use CI tool like Jenkins for automatic build and deployment

Configuration consistency for large number of nodes

»        Automatic validation of configurations before and after the execution

»        Take restore point and roll it back after the performance run

Population of high volume of test data

»        Copy production data and mask it

»        Alteration of DB volumes

»        Use tool like database benerator, dbmonster

»        Use historic data

Simulation of realistic load

»        Identify key scenarios and usage patterns from log files, market research, BA etc.

»        Generate load from different geographies

»        Baseline response with CDN and without CDN

Penetrating the system complexity, touching all system nodes and database tables

»        Understand system architecture

»        Manually walkthrough the scenarios and watch traffic on different nodes and database tables

»        Detail analysis of application logs

»        Understand load balancer strategy

Identification and testing of failover scenario

»        Test the failover scenario separately during load condition

Third party interactions

»        Simulate using stubs

Monitoring of large number of disparate systems

»        Use diagnostic tools like AppDynamics, DynaTrace, HP Diagnostic, Glassbox

Result collation and analysis

»        Automatic result collection and collation

»        Collection of built-in anti patterns for quick analysis

»        Knowledge base on historical failures or bottlenecks

 

Comments

Post a Comment

Popular posts from this blog

Performance Test Run Report Template

You have gathered requirements from clients, identified key critical business scenarios, analyzed the workload distributions, found out performance goals and requirements, identified  tool for generating load, setup the test environment, designed the scripts, executed the performance run, collected the raw data, analyzed it, what next?  Send information to stakeholders? No!!  Wait!! It needs to formatted, customized for different stakeholder(s) and reported accordingly. This article suggests one of the ways to report performance run results to client especially if the performance testing activity has been outsourced to yours company.  It discusses the performance report template and importantly the reasons for including the items in the report. The performance report template contains information by data collected from client side only and doesn’t discuss / talk about data from server side like processor utilization etc. Broadly the performance report has been divided in three major
Read more

Bugs Management in Agile Project

Agile has a refreshing approach to bugs, especially when contrasted with traditional waterfall practices. Historically, teams would postpone bugs to a stabilization phase. This stabilization phase was long and tedious, typically with high stress and long hours. Agile, on the other hand, promotes the following thought processes: Carrying bug debt is unhealthy and sometimes, even destructive. We should fix bugs, not track them. A feature isn't done until the bugs are fixed. Quality is part of the overall cost. Bugs are prioritized against new features. A bug needs to be more important than the next new feature. Notes All new features have a "bug tail." The bug tail is the diminishing number of open bugs on the new functionality. When a feature is added, bugs are introduced. Most are discovered and fixed immediately by the team as a part of feature development: some a few days later, others not until long after a story is considered &quo
Read more

Understanding Blockchain

This notes will cover below questions... Introduction to Blockchain What is the difference between Blockchain and Bitcoin? What are some other uses of blockchain that go beyond bitcoin? History of blockchain Could you explain the blockchain process in laymen terms? If all data is available to the public. Can anyone know my bank balance (in Bitcoin)? Introduction to Blockchain The Blockchain is a decentralized and distributed public (digital) ledger where transactions are recorded and confirmed anonymously. It’s a record of events that is shared between many parties. More importantly, once information is entered, it cannot be altered without altering subsequent blocks. WTF is the blockchain? A guide for total beginners What is the difference between Blockchain and Bitcoin? The blockchain is a technology
Read more