load runner 80

8/8/2019 Load Runner 80

http://slidepdf.com/reader/full/load-runner-80 1/116

LoadRunner Quick Tutorial

LR78-SLIDES-01A



第一节

LoadRunner介绍



为什么要进行压力测试

The failure of a mission-critical application can be costly

Assure performance and functionality under real-world conditions

Locate potential problems before your customers do

Reduce development time

Reduce infrastructure costs



压力测试分类

Component Testing

Find the behavior and performance of each tier.

Load Testing Find out whether the system can handle the

expected load upon deployment under real-world

conditions.

Find the application's breaking point. Apply testing

that measures whether the application's

environment is properly configured to handle

expected or potentially unexpected high transaction

Stress Testing

volumes.

Volume Testing

Find the stability of the system with respect to

handling large amounts of data over extended

time periods.



压力测试目标举例

Application response time

How long does it take to complete a

task?

Configuration sizing

Which configuration provides the

best performance level?

Acceptance

Is the system stable enough to go

into production?

Regression

Does the new version of the software

adversely affect response time?

Reliability

How stable is the system under a

heavy work load?

Capacity planning

At what point does degradation in

performance occur?

Bottleneck identification

What is the cause of degradation in

performance?

Product evaluation

What is the best server for 100

users?



手工测试的问题

Testers

Do you have the testing resources?• Testing personnel• Client machines

How do you synchronize users?

How do you collect and analyze results?

How do you achieve test repeatability?

Everybody,click theSEARCH

button again

Coordinator

Applicationserver

Databaseserver

Load Generation System Under Test



LoadRunner解决方案

Overcomes resource limitations

• Provides meaningful results with

analysis tools

• Allows repeatable tests with scriptedactions

Analysis

• Runs many Vusers on few machines

• Replaces testers with "Virtual Users"

• Controller manages Virtual Users

Databaseserver

Controller

Vusers

VirtualUser loadgenerator

Applicationserver

Load Generation System Under Test



LoadRunner的实现方法

LoadRunner is a complete solution to conduct automated load

testing before, during, and after application deployment.

LoadRunner allows you to...

Scale

the application to

a predetermined

capacity.

Identify

business processes

or system

components havingtrouble.

Diagnose

a problem to

pinpoint the

root cause.



LoadRunner 组件

Vuser Script Recorder – records Vuser scripts thatsimulate the steps of real users using the application.

LRVUGEN

The Controller is an administrative center for

creating, maintaining and executing Scenarios. The

Controller assigns Vusers and load generators to

Scenarios, starts and stops load tests and performs

other administrative tasks.

LRCONTROLLER

LR Analysis provides graphs and reports. These graphs

and reports summarize the system performance.LRANALYSIS

This course focuses on LR Controller and LR Analysis. LR VuGen is discussedin a separate course.



LoadRunner 支持的环境

Platforms:NT, 2000,XPSunHP

IBMLinux

Web:HTTP(S)CORBAWAPCOM

RMILDAPWinsockStreamingFTP

IIOPPOP3iModeRealPlayerMS Media

Voice XMLSMTP

MiddlewareProtocols:CORBACOMEJBs

TUXEDOJoltJDBCMQSeries

ERP/CRM:Oracle Apps.SAPSiebelPeopleSoft

ClarifyBaan

Databases:OracleMS SQLServerDB2Sybase

InformixODBC

Legacy:32705250VT100

In ternet / In t ranet Webservers Appservers Databaseserver Cl ients



什么是 Scenario?

Scenario

A scenario is a file that defines which

Vusers to execute, the number of

Vusers to run, the goals of the test,

which computer hosts the Vusers and

the conditions under which to run the

load test.scripts

run-timesettings

loadgenerators

real-timemonitors

virtual users

Elements of a Scenario



LoadRunner的工作方式

Databaseserver

Appservers

Webservers

LRCONTROLLER

LOADGENERATOR 2

(NY)

LRANALYSIS

LOADGENERATOR 1(SF)

1.Create scenarios

and assign them to

load generators.

2.Generates load and

sends performance

data to controller.

3. Access the reports

and graphs.

2.Generates

load and sends

performance

data to

controller.

Performance metrics



硬件配置需求

Win NT

Win 2000Win XP

500 MB

Win NT

Win 2000Win XP

UNIX

Depends on

the number

of Vusers and

runtime

settings.

Win NT

Win 2000Win XP

200 MB

Operating System

Free Hard Disk

Space

256 MB256 MB256 MBMemory

AnalysisLoad

Generator

ControllerLR

Components



压力测试流程

TweakSystem Based

on Analysis

Plan Load

Test

Plan Load

Test

1

Create

Virtual Users

Create

Virtual Users

2

Scenario

Creation

Scenario

Creation

3

Scenario

Execution

Scenario

Execution

4

Results

Analysis

Results

Analysis

5

CreateScenarios

based onyour goals

Fine tunesystem

based onanalysisdata andre-run thescenario

DefineGoals

Analyzesystem

GatherSystem

UsageData

RecordScripts and

turn scriptsinto Vusers byadding testdata and userbehavior

ExecuteScenarios

for testingfirst,running100% andoverloading

Analyzeresults

data andpinpointbottlenecks

Rerunning the scenario indicates whetherperformance has improved and whether goals

have been achieved.



压力测试是一项团队工作

Manages the test applicationimplementation for all phases of

load testing

Provides expertise on the businessfunctionality of the applicationunder test for planning phase

Team Member Main Responsibility

Project Manager

Functional Expert

Creates the multi-user load onthe system for all phases of loadtesting

LoadRunner Expert

Analyzes and tunes the

application server for resultsanalysis phase

Application Server

Performance BasisExpert

Analyzes and tunes the databaseserver for results analysis phase

Database PerformanceBasis Expert



LoadRunner 团队工作

Tweak

System Basedon Analysis

Plan Load

Test

Plan Load

Test

1

Create

Virtual Users

Create

Virtual Users

2

Scenario

Creation

Scenario

Creation

3

Scenario

Execution

Scenario

Execution

4

Results

Analysis

Results

Analysis

5

LR Expert

FunctionalExpert

ProjectManager

LR Expert

FunctionalExpert

ProjectManager

LR Expert LR Expert

Sys Admin

ServerExpert(s)

Sys Admin

ServerExpert(s)

LR Expert



第二节

规划



在创建脚本之前要进行充分的规划

Plan Load

Test

Plan Load

Test

1

Create

Virtual Users

Create

Virtual Users

2

Scenario

Creation

Scenario

Creation

3

Scenario

Execution

Scenario

Execution

4

Results

Analysis

Results

Analysis

5

Goals, business processes,

user type, user behaviors, input

data, system profiling,

scenario design, and LoadRunnerinstallation



规划阶段的内容

1. 建立负载测试的目标制定出可以用数量衡量的目标标准

2. 分析被测系统的业务运行状况

确定要录制哪些商业流程(business processes)

确定业务高峰的负载情况和时间

把需要测试的BP的操作步骤形成正式文档

准备测试数据

3. 收集系统的技术信息

确定系统的架构确定系统各部件的软件情况，使用什么样的数据库服务器？

Web服务器？以帮助我们在合理的时间内确定系统瓶颈



本次培训使用的Demo软件

St ar t > Programs > Mercury Tours

Before you access the

training application:

Invoke the Apache

Server (Start >Programs > Start

Server)



第三节

编写脚本



录制Web脚本的方法

1.运行VuGen，创建新的脚本

2.选择创建Web(HTTP/HTML) protocol的脚本

3.设定录制选项 Recording Options.

4.开始录制

5.进行正常的业务流程的操作

6.当业务操作完毕后，按Stop Recording将录制结束

7.保存脚本

DEMO – 老师将会演示如何录制脚本



设定录制选项 Recording Options

Recording

Browser

Recording Proxy

Advanced

Correlation

DEMO – Recording Options



两种不同查看脚本的方式

Script ViewTree View

Each Tree View icon

represents a ‘C’ statementviewable in Script View

Tree view has icons for the

recorded user steps

DEMO – Tree and Script View

– Icon and usage examples



什么时候选择不同的录制方式？

HTML

Recording:This is default mode.

If problem encountered with form submission or

with link clicking then record with URL optionchecked under HTML Advanced

Records a separate request for each objectrequested from the server. Gives best control

over traffic.

URLRecording:



脚本的目录结构

C:\ C:\

web_projectweb_project-

ScriptsScripts-

-

Action1.cAction1.c

vuser_init.cvuser_init.c

vuser_end.cvuser_end.c

Lab1Script2.usrLab1Script2.usr

As soon as the script is

saved, these are the

folders that you will see

Lab1Script2Lab1Script2



设置回放的运行时刻设置Run-time Settings

Run Logic

Pacing

Log

Think Time

Miscellaneous

Speed Simulation

Browser Emulation

Proxy

Preferences

ContentCheck

Available for

all protocols.

Specific to

Web.

DEMO – Run-time Settings



运行脚本

Click the Run

button to play

back the script.

The Execution loglists all messages

once the script has

stopped running.

DEMO – Running a script and how to set results directory



常用的调试脚本的工具

Test Results window

Execution Log

Snapshot Results window

Page View

Server Response

Client Request

Run-Time Viewer

DEMO – Available debugging tool options



本节总结

本节介绍了常用的录制选项

本节介绍了两种录制Web协议脚本的方法

本节介绍了运行时刻的各种设定

本节介绍了一般选项的设定

本节介绍了进行调试脚本的基本工具



什么是Action?

Action

A set of recorded user actions in a

Vuser that allows you to emulate a

particular user task through script

iteration.



在Web脚本里面使用多个Action

A Web Vuser (script):

Search for a CD Order the CD

Log in

Go to checkout

vuser_init

Run once

Action

Run 3 times

vuser_end

Run once

Search for a DVD Order the DVD Action1Run 5 times

Search for a book Order the book Action2Run 4 times

then

then

then

then

Action



使用多个 Action的典型例子

最典型的情况是用户登录一次，进行多次业务，然后退出系统:

Record Login user steps in ‘vuser_init’

Record selecting an item in the first section called ‘Action’

Record placing items in shopping cart user steps in ‘Action1’

Record Checkout and Logout user steps in ‘vuser_end’

Internet

DBserver

Webserver



如何把不同的操作录制到多个 Actions中?

Action list

1. Start recording

2. Select action file

3. Record users steps in

the appropriate

action file

4. Repeat steps 2 and 3as necessary

5. Stop recording

DEMO – Adding creating LR Actions



注意设定好开始和结束的状态

Always end at this page

if the next iteration is

Sign-On or Register.



根据测试目标来确定脚本的Action

Emulate three types of users where 50% of the users are logging

in, 25% are searching and the remaining 25% are booking tickets?

Load Testing Goal:

Suggested method:Record one script with all three business processes under

separate actions and apply weighting.

DEMO – Creating Weighting



如何衡量操作的响应时间？

Start ApplicationSTEP 1

How can I isolate

a measurement of

just the Sign-in

response time?

Type username and passwordSTEP 2

Click on Sign-In buttonSTEP 3

Wait for confirmationSTEP 4

Select a departure and

arrival citiesSTEP 5

Enter datesSTEP 6

STEP 7 Enter credit card details

Click on SubmitSTEP 8



什么是Transaction?

LoadRunner TransactionGives an end-to-end measurement of time

elapsed in the execution of one or more steps

in a business process.



LoadRunner的Transactions测量什么内容?

Request

(Transaction starts)

Sign-In

transaction

took 8 secResponse

(Transaction ends)

Web

ServerUser

LoadRunner transactions measure the transmission

and response time between client and server.

Transaction



Transaction开始和结束按钮

Common icons and

usage examples

Start Transaction Set starting point for

measurement

End Transaction Set end point for

measurement



Transaction的状态

LR_AUTO – (default) - use this to detect the transaction status

automatically

LR_PASS / LR_FAIL / LR_STOP – use these when you

want to set the transaction status manually in the

code

Transactions



Transactions中的思考时间如何处理？

Think times May reduce accuracy of

transaction measurements

Can be filtered out of Analysis

graphs

通过检查运行日志来确定



Transaction的执行结果

Transactions



Transactions的相关设定



参数化的概念

Recorded Script

Select San Francisco

Selection is hard coded and willselect San Francisco on every run

Parameterized Script

Select <dep_city>

Selection is parameterized and willselect different city on every run

“ ”?



为什么要引入“参数化”?

REAL USERS

Varied input to

the application

Static input to

the application

Recorded

script

(Virtual Users)

To Emulate real-world behavior and to create realistic results



什么时候需要参数化？

Start ApplicationSTEP 1

Type username and passwordSTEP 2

Datadependency

STEP 3 Click on Sign-In button

STEP 4

Unique

constraint Wait for confirmation

Select a departure and

arrival citiesSTEP 5

Data

cachingEnter datesSTEP 6

STEP 7 Enter credit card details

STEP 8

Date

constraint

Click on Submit



如何参数化输入数据

1.Determine which fields to parameterize.

2.Replace recorded values with parameters.

3.Decide which parameter type to use.

4.Create the data file.

5.Populate the data file.

6.Choose the data access method and the number of

iterations.7.Run the Vuser and analyze the results to verify correct

execution.

DEMO – How to parameterize a input data



参数的类型

Specifying the source for the parameter data.

There are three types:

Internal Data

Data Files

Master Data

User Generated Data

External Data

User-Defined Functions

‘File’类型参数的数据访问方式



• Sequential

• Random

• Unique• Same line as

<parameter>

Select next tells LoadRunner in what order to use data

from the file (data source)

如何查看脚本执行时参数的值？



To see parameter data in the Execution Log, select Ext ended log and Paramet er subst i t ut ion in the Run-time Settings

数据个数和Run Logic的关系



To test all data using the Sequential access method,

Number of Iterations should equal the number of rows in the data file



什么是检查点Checkpoint?

CheckpointA checkpoint is a step in a Vuser script

which verifies whether a visual cue

appeared as expected.

Verification always results in a PASS or FAIL status

在大并发用户压力的情况下验证脚本是否执



行成功是负载测试的基本要求

Success. Business

process complete.Low load

One Vuser running

the script

Failure. Business

process failed.Increase in load

Script run under load withmany concurrent Vusers

确定脚本执行成功的标准



What indicates theticket was purchased

successfully?

Login

Select departureand arrival cities

Select departureand arrival dates

Select flight

Purchase ticket

ContinueShopping

There was an errorwhile generating your

order!

Your itinerary wasbooked! Thank you forchoosing MercuryTours.

通过界面的某些信息来确定脚本是否执行成功



Flight confirmation

image and text appearafter pressing the

Secure Purchase button.

Image Confirmation

Text Confirmation



检查点的基本内容

(1) Capture the information needed to performverification.

(2) On playback, compare the actual value on theapplication to the saved expected value.

(3) Send the comparison result (the "status") as either

PASS – Test requirement # 5 successful.OR

FAIL – Test requirement # 5 failed.

ExpectedValue

ActualValue

PASSIF

LoadRunner



LoadRunner提供的文本检查点

VuGen allows you to add Checkpoints for verification

automatically and also programmatically.

Checkpoint Icons

Text A text check is a search for a

specified text string on a Web page.

Adds web_reg_find function tothe script.

关联的基本概念介绍 – Recording



VuGen Script

Login

(User Name jojo Password bean )

1.Log in

User Name jojo

Password bean

2.Your sessionid@@@12345

3.Request for product

info

4.Product info found

Response

(SessionID@@@12345 )

Request for product info


Web

Server

Response

Product info

Dynamic data is hard-coded in the script

关联的基本概念介绍 - Playing Back



VuGen Script

Login

(User Name jojo

Password bean )

1.Login

User Name jojo

Password bean

2.Your sessionid@@@12345

3.Request for productinfo

4.Product info found

Response


Web

ServerRequest for product info


Response

ERROR SessionID 12345 has expired

Login again to get product info

Reusing recorded dynamic data will cause errors when you replay the script

关联的基本概念介绍 - Solution



Web

Server

Login

(User Name jojo

Password bean )

Response


Request for product info

(<SessionID@@@23456>)

Response

Here is the requested product info

Correlat ed VuGen Script

1. ……

2.Your <session_id>

3. ……

4. ………

Recorded dynamic data must be correlated using parameters



关联是一个原理性的问题

关联是一个原理性的问题，在任何协议的脚本中都存在1

2

关联是开发脚本的最耗时，最烦琐的一个过程.因为每个被测系统都不同

3为了减轻关联的工作量，LoadRunner提供了一些自动化关联的辅助工具



手工关联的基本步骤

1. Play back the script and determine if error is due tocorrelation.

2. Determine which dynamic values to correlate.

3. Find the left boundary, right boundary and occurrence of thedynamic value.

4. Add a web_reg_save_param function and the arguments tothe script, above the step which requests the page with thevalue to capture.

5. Parameterize the dynamic value in the script every time itoccurs.

6. Verify correct execution.

关联的第一步：脚本回放失败



Q: Why did the step creation fail?

1. Make a note of which step failed.

2. Investigate the Execution Log.

通过WDiff 查看脚本变化的内容



There are five differences highlighted.

确定哪些动态信息需要关联• These values represent the screen locations of mouse clicks



ese a ues ep ese t t e sc ee ocat o s o ouse c c s

• They are used by the script but do not need to be correlated

"Name=buyFlights.x", "Value=52", ENDITEM, "Name=buyFlights.x", "Value=23", ENDITEM,

"Name=buyFlights.y", "Value=15", ENDITEM, "Name=buyFlights.y", "Value=17", ENDITEM,

lr_think_time( 2 ); lr_think_time( 1 );

lr_think_time( 7 ); lr_think_time( 11 );

"Name=flight_210", "Value=on", ENDITEM, "Name=flight_206", "Value=on", ENDITEM,

"Name=removeSelected.x", "Value=101", ENDITEM, "Name=removeSelected.x", "Value=80", ENDITEM,

"Name=removeSelected.y", "Value=13", ENDITEM, "Name=removeSelected.y", "Value=10", ENDITEM,

lr_think_time( 1 );

1

2

3

4

5

Think times are used by the scriptbut do not need to be correlated

Flight number is dynamic and

requires correlation

关联的函数: web_reg_save_param



web_reg_save_param has the following arguments:

User-defined parameter name Indicates "no more attributes"

("param_name", <List of Attributes>*, LAST);

"LB=", "RB=", "ORD=",

Optional indicatesordinal or instance orthe match

The left & right boundaryof the parameter or thedynamic data

*See the LoadRunner documentation for a complete list of attributes

在Server Response中确定动态信息的左右边界



LEFT BOUNDARY

RIGHT BOUNDARY

6

LB: type="checkbox" name=" RB:"

脚本录制完成后自动关联的方法



需要将脚本回放一次

LoadRunner会自动比较回放时的结果和录制时的结果，确定需

要关联的内容

回放完成后可以使用LoadRunner内置的Scan f or Correlat ions

工具进行自动关联

脚本录制时自动关联的步骤



1. Enable Correlation during recording in Recording Options2. Create new application

3. Create new rule for the newly created application anddefine properties for the new rule

4. Test the newly created rule (optional)

5. Record the script

6. Play back the script to verify correct execution

DEMO – Correlation During Recording

第四节



Scenario设计

目标



Understanding Scenarios

Scheduling Scenarios

Runtime Settings

Performance Monitors

Tweak

System Basedon Analysis

Plan Load

Test

Plan Load

Test

1

Create

Virtual Users

Create

Virtual Users

2

Scenario

Creation

Scenario

Creation

3

Scenario

Execution

Scenario

Execution

4

Results

Analysis

Results

Analysis

5

什么是 Scenario?



Scenario

A scenario is a file that defines which

Vusers to execute, the number of

Vusers to run, the goals of the test,

which computer hosts the Vusers and

the conditions under which to run the

load test.scripts

run-timesettings

loadgenerators

real-timemonitors

virtual users

Elements of a Scenario

LoadRunner Controller

Before scenario execution



Before scenario execution

used to design scenario

used to initiate scenario run

During scenario execution runs many Vusers simultaneously

controls each Vuser (initialize, run,

pause, stop)

displays execution status of each Vuser displays messages from each Vuser

monitors system and network resources

After scenario execution collects and organizes performance data

launches the Analysis tools

Controller

organizes

and manages

scenario

elements

DEMO – Controller

选择 Scenario 类型



Choose one of

these types when you start a

new scenario:

Manual

Goal-Oriented

DEMO – Scenario Types

Manual 与 Goal-Oriented Scenarios 比较



Manual Scenario

main purpose is to learn how many Vusers can

run concurrently gives you manual control over how many Vusers

run and at what times

Goal-Oriented Scenario

goal may be throughput, response time or

number of concurrent Vusers

LoadRunner manages Vusers automatically

选择哪一种 Scenarios 类型?



Scenario Outline Scenario Type

1.Script should define ‘Update’

2.When running the load test at peakload, achieve 1,000 concurrent users

Manual scenario with

1,000 concurrentVusers

1. Script should define ‘Search’

Transaction

2. When running the load test during non-

peak hours achieve response time of 8

secs. with 2,000 concurrent users

3. When running the load test in peak hours

achieve response time of 12 secs with

5,000 concurrent users

Goal-Oriented scenario

with Transaction Response Time as the

‘Goal Type’

Scheduling Scenarios



Allows repetition of important usage patterns throughout

load testing for defined duration of load test.

DEMO – How to access Schedule Builder

Schedule by Scenario

Applies same schedule across all groups



Allows delay in start time for scenario

Creates a profile of usage by “users”

Schedule by Group



Individual settings for each group

Allows delay in start time for scenario

Creates a profile of usage by “business process”

Ramp Up 与 Ramp Down



Ramp Up Allows loading all Vusers in the scenario gradually

Timed increments pinpoint the exact point at which

a system/software fails or degrades

Ramp Down

Only applicable for scenarios where in theDuration tab Run for option is selected

Specify how many Vusers to stop and how often

Stop all Vusers at once

Primarily used to detect memory leak and system

recovery after heavy load

什么是 Run-time Setting



Run-time Settings can be configured in VuGen and in

the Controller

Different Groups of Vusers many have different Run-timeSettings

Run-time Settings greatly affect Vuser behavior

Run-time Settings greatly affect Vuser “footprints”

Run-time Settings



Run Logic

Pacing

Log

Think-Time

Miscellaneous

DEMO – How to access Run-time Settings

Run Logic 与 Pacing

Used to model more closely the way real users interact



y y

with the application

Run Logic

Replicates a real user performing steps in a business processmore than once

Allows setting of the iteration count (how many times a Vuser

will repeat a task)

Pacing

Controls the delay between iterations

Can be used to extend the amount of time a test runs

Log 设置



In Controller, logging is enabled by default to send messages

only when there is an error

Used for Vuser information such as

error messages

parameter substitution

transaction status verification status

Think Time



Think Time introduces the right amount of delay to

emulate user behavior accurately

Allows replaying or ignoring of recorded Think Time(default = replay)

Influences how the Vuser uses the recorded Think Time

Ignoring Think Time will greatly affect the scalability of the

load generators

Miscellaneous Settings

Error Handling



Continue on Error allows Vusers to attempt to proceed past an error.

It is used when Error Handling has been added to the script manually.

By default, Continue on Error is off.

Multithreading

Threading supports more Vusers per load generator

Process supports Vuser than can not be multithreaded

Automatic Transactions

Actions can be automatically “timed” as a Transaction. Script Actions

should be assigned meaningful names for clarity if automatic

transactions are used.

Step can also be automatically “timed” as a Transaction. Script steps

should be assigned meaningful names for clarity if automatic

transactions are used.

Scenario Run-time Settings

Scenario Run-time Settings relate to:



Vuser QuotasAllows setting the number of Vusers to initialize at one time.

Applies to all load generators. This setting can apply to

individual load generators.

Stopping VusersAllows setting how LoadRunner will stop the running of Vusers.

Random Sequence SeedEach seed value represents one sequence of random values

used for test execution. This setting applies parameterized

Vuser scripts and randomized think times.

DEMO – How to access Scenario Run-time Settings

Transactions 不能说明一切

Transactions measure the time required for a business

Th f f t ti t



process. The performance of transactions suggests apicture of the system component’s behavior.

Transaction: search

Exercises the Web, applicationand database server

This transaction performed

poorly with a response time of 30 seconds.

Transaction: buyExercises the Web and

application server

This transaction performed

as expected with a response

time of 1 min.


Apart from Transactions performance, you need Performance monitors

to highlight system component bottlenecks



to highlight system component bottlenecks

In ternet / In t ranetDatabaseserver

Appservers

WebserversCl ients

LoadRunner offers wide range of performance monitors for isolatingbottlenecks

Monitors are non-intrusive and agent-less Monitors gather data for online/offline analysis

Monitors display real-time data during testing

Monitors gather data for offline root cause analysis after testing




View and collect statistics during Scenario execution

using LoadRunner Controller online graphs

Correlate the collected statistical data in LoadRunnerAnalysis after scenario execution for root cause analysis

Facilitate collaboration between project members

during and after Scenario execution by allowing users toview multiple performance monitors via the browser

DEMO – Run Scenario with BookFlight script and how

online monitors give the current statistics.

- How to add and delete monitors.

配置 Monitor

To add measurement for

System Resource monitor



System Resource monitor.Do this:

1. Right-click a monitor pane

2. Select "Add Measure-

ment(s)" from pop-up

menu

3. Select server machine

name and resources to

monitor from the dialog

Administrative privileges are required to configure monitors

DEMO – How to configure monitors.

Monitor Measurements

Many system level



Many system-level

resources can be

measured. By

default, LoadRunner

selects many of the

important

measurements.

Consult with subjectmatter experts to

determine what

else should be

monitored.Measurements

added for Windows

Resources monitor

第五节



Scenario Execution

执行 Scenario 之前的准备工作

Add performance monitors in Run tab



Add performance monitors in Run tab.

Specify a location to save the scenario results and decide on a

name for the results.

Establish rendezvous policies, if any.

Executing Scenario 过程中的团队合作

Load test engineer monitors

transaction performance



Database

server

App

servers

Web

servers

transaction performanceAnd server monitors via

LoadRunner Controller

Network Administratormonitors network

performance while

load test is running

Application or DatabaseAdministrator monitors

remote system

performance

during load test

Scenario Execution 过程: Debug Run



DEBUG RUN

3-5 Vusers

ISOLATE

TOP TIME

TRANSACTIONS

20% LOAD

SCALABILITY

100% ++

LOAD

FULL LOAD

100%

Objective: Verify that parameterized data works under concurrent usage

Run-time Settings: Use extended log and data returned by server and

ignore think times

What to look for: No errors. This signifies that you are now ready toproceed with the load test

Scenario Execution过程: Isolate Top TimeTransactions



SCALABILITY

100% ++

LOAD

FULL LOAD

100%

DEBUG RUN

3-5 Vusers

ISOLATE

TOP TIME

TRANSACTIONS

20% LOAD

Objective: Isolate transactions that have performance problems

irrespective of load to find if there are any bottlenecks

Run-time Settings: Use standard log and ignore think time

What to look for: The worst performing transactions. This might be inabsolute time (longest) or it might be against

some measurable goal (2X as slow as the test objective)

Scenario Execution 过程: Full Load

SO



SCALABILITY

100% ++

LOAD

DEBUG RUN

3-5 Vusers

ISOLATE

TOP TIME

TRANSACTIONS

20% LOAD

FULL LOAD

100%

Objective: Verify that the system performs as expected under load

Run-time Settings: Turn logging off and enable think time

What to look for: Compare your load test goals to the test results

Scenario Execution过程: Scalability Test

ISOLATE



DEBUG RUN

3-5 Vusers

ISOLATE

TOP TIME

TRANSACTIONS

20% LOAD

FULL LOAD

100%

SCALABILITY

100% ++

LOAD

Objective: Verifies the load limit that can be handled before more

resources are required

Run-time Settings: Turn logging off and enable think time

What to look for: Load test goal should be exceeded, and 2X responsetime of your full load test (100%) should be achieved

Vuser 状态

Initialize and Run buttons affect selected Vusers.



Initialize and Run buttons affect selected Vusers.

Start Scenario runs all Vusers.

Vusers button opens Vusers window and shows status

of Vusers.

Vuser Window 与 Vuser Log



Scenario 状态

The Scenario Status window allows you to view:N b f V tl i



The Scenario Status window allows you to view: Number of Vusers currently running

Elapsed time

Hits/Second Passed/Failed transactions

Error count

Performance Monitor 状态

Start by looking at system resources – CPU

consumption and memory usage of servers.



p y g Then start observing application resources – are

there any 404 or 500 errors?

Use the above information in combination withresponse time and number of Vusers.

第六节



Result Analysis

哪里出现了问题 ?

Any link in the chain between browser and

database could be the cause of an applicationf il f d d ti



failure or performance degradation.

Appservers

Database

servers

Webserver

Web server slowdown

DB server slowdown

Network slowdown

Web

Client

Root Cause Analysis



Component metrics

System-level performance

Network and Server Response Times

End-user experienceTransaction Response Times

Performing root

cause analysis

means drillingdown from the

most general

reports to localized

metrics.

达到测试目标了吗 ?

The Summary Report provides a high-level view of the results of our

test



Max response

times

11 transactionsfailed

Average Transaction Response Time

Details transactionresponse throughout



response throughout

the test

Identifies

problematic

transactions

Identifies

problematicpoints in the test

Average Transaction Response Time

Under Load

Details transactionresponse time as a



response time as a

function of load

Identifies under what

load conditions

transaction times

begin to degrade

Transaction Response Time Percentile

Details transactionresponse time as a



response time as a

service level

Identifies

percentage of

transactions

complete over

time

Transaction Response Time Distribution

Detailstransaction



response time as

a distribution

count

Identifies number

of transactions

complete over

time

数据分析

The amount of data that can be gathered during a test is

staggering. Without some effective techniques for

i l i hi d di i i ld b



manipulating this data, diagnosing issues would be

impossible. LR Analysis provides:

Filtering to hide unwanted data from analysis

Regrouping to assemble data as desired by merging graphs,

tiling and auto-correlation

Duplicate graphing to separate data

Legend Filtering to update graph displays quickly

Drill-Down

DEMO – Filtering, Legend Filtering and Drill Down.

使用 Auto Correlation 查找性能瓶颈

Auto Correlation

Allows automatic selection of graphs to correlate in



order find the metrics that most influence thetransaction response time

Automatically correlates the transaction graph withALL performance metrics and displays the ones thatshow the highest match in their behavior; this allowsyou to disregard irrelevant metrics

Visually displays correlated metrics closelycorrelated to the transaction response time, whichspeeds identification of the likely causes of

performance bottlenecks

DEMO – Auto Correlation.

Analysis Reports

HTML FormatEase of sharing analysis results with other remote team



members

Word FormatCan add comments and format the analysis data based

on your needs

load runner 80

Documents