Giter Club home page Giter Club logo

tpc-w's Introduction

Overview

TPC-W is a transactional web e-Commerce benchmark specified by the Transaction Processing Performance Council.

We started with an open-source Java implementation available online, that already runs on top of Tomcat and MySQL. The business operations are implemented using hardcoded SQL queries to the database. Initially, we adapted it to run on top of the Fénix Framework. This adaptation required transforming the hardcoded SQL queries into Java code. The detailed process is documented here.

The domain model

Using Fénix Framework, the TPC-W domain model is described using the Domain Modeling Language (DML). DML is a domain-specific language that enables programmers to specify the structure of an application's domain model in a Java-like syntax, and then automatically generate the structural Java code for the classes. The advantage is that it is shorter to write, and it also represents relations between domain classes as a first-class concept. Furthermore, the DML compiler can generate backend-specific code that maintains the same programming API on top of the domain entities. This allows us to replace backends without having to change the application.

Installation

In the following we describe how to install and run the TPC-W Benchmark. This benchmark is composed by two separate components: the application server and the browser emulator. The former is deployed to a web server and the latter runs a given number of threads that simulate concurrent web clients interacting with the application.

Requirements

To run the benchmark:

  • Apache Tomcat 6.x (should work on 7+ versions but hasn't been tested)
  • JDK >= 1.5

To plot the measurement results:

  • FreeMat

Get the source code

git clone git://github.com/fenix-framework/examples.git

In the examples/tpcw directory there are two independent programs in server and client. The server is the web application to be deployed in Tomcat. The client contains the browser emulator. The client also contains some FreeMat files that process the benchmark's output, and the original TPC-W implementation for reference. From now on, operations regarding each program should be performed in that program's top directory (e.g. to perform the instructions regarding the server go to examples/tpcw/server).

Running the benchmark

This is the quick-and-dirty how to get this benchmark running. For other alternatives please take a look at the configuration files within the source code.

Required configuration for Tomcat

You should configure in your Tomcat a user that has permission to deploy a web application, and then take note of those credentials. Then start the Tomcat web application server.

Build and deploy the application server

To create the WAR packaged application do:

mvn package

You may change the backend to another by passing the property -Dfenixframework.code.generator in the previous command. Have a look at the pom.xml file for some possible values.

You may also change the following parameters when creating your application:

mvn package -DNUM_ITEMS=1000 -DNUM_EBS=10

The values shown are the default. They represent:

  • NUM_ITEMS: the maximum number of books in the database and it must be one of 10^b with b in [3;7].

  • NUM_EBS: the maximum number of emulated browsers that TPC-W will need to support. It must be greater than 0.

If the package command succeeded you should be able to deploy the application with:

mvn tomcat6:deploy -Dtomcat.username=<username> -Dtomcat.password=<password>

Replace <username> and <password> with the correct credentials for your Tomcat installation.

NOTE: Be careful in changing the default values, as they have an implication in the number of instances of many domain objects and increasing them can easily grow the dataset out of manageable proportions! It is suggested that you start with the default values.

Also, if later you wish to replace the deployed application server with another version make sure you first remove the previously deployed code with:

mvn tomcat6:undeploy -Dtomcat.username=<username> -Dtomcat.password=<password>    

Check that the application is successfully deployed

Go to TPC-W's home page and you should see a page similar to this. Images in the loaded page do not show. This is not an error. It occurs simply because we didn't package the site's images in the web application.

You SHOULD NOT click any other links yet, because the data set is not populated. This was just to check whether the application is successfully deployed.

Populate the dataset

Given that the default configuration uses Infinispan as an in-memory-only data repository, we need to generate a data set every time we redeploy the server application (which embeds Infinispan). To do so we've created a servlet that can be activated with:

wget -q -T 0 -O - "http://localhost:8080/tpcw/TPCW_populate?NUM_EBS=10&NUM_ITEMS=1000"

In here the values for NUM_EBS and NUM_ITEMS should always be less than or equal to the values used when building the application. Again, the parameters shown in the example are the default values, so they could be omitted.

If you do not have wget installed on your system you can alternatively access the given URL in your browser. Just make sure you NEVER reload the page. It may take a while to populate the data. You must wait for it to finish before proceeding. The population servlet is meant to run only once and it does not support multiple executions.

Note: You may want to check in the server's logs (${CATALINA_HOME}/logs/catalina.out) that the data population was successful.

Build the test client

Now move to the client application and run:

ant dist-client-only

This creates the client application in the dist directory along with a shell script to run it.

Run the client

Just do something like:

cd dist
chmod 700 rbe.sh
./rbe.sh -r demo -u 60  -i 120 -d 20 -t 1 -b 1000 -n 3 -tt 0

This simulates 3 store clients continuously accessing the store during 120 seconds, with a previous warm-up period of 60 seconds and a ramp-down period of 20 seconds, using workload 1 (TPC-W's browsing mix with 5% write transactions). To better understand the parameters and the values that can be used run:

./rbe.sh -h

Just take note that the value of flag -b must match the value of NUM_ITEMS used previously.

Analyze the results

After the script completes a FreeMat file is produced with a name like rundemo_t1_e3_b1000_20111011_2301.m. This file must be processed in FreeMat. To do so you'll need the files in directory freemat. Just copy the generated file over tpcw.m and then open FreeMat.

cp rundemo_t1_e3_b1000_20111011_2301.m ../freemat/tpcw.m

In FreeMat you will probably need to configure its path to point to the freemat directory in the client. Running the function wips(tpcw) will produce a plot with the Web Interactions Per Second (WIPS) corresponding to the current contents of tpcw.m. The plot shows a timeline and the WIPS for every second of the execution. It also plots a curve with the average WIPS for the 30 seconds around a given point, and a line with the overall average WIPS.

This is an example of the kind of FreeMat plot you can expect to see.

A sample script to automate stuff

The following is an example script for running a sequence of tests. You most probably will need to adjust it to match your needs. When invoking the script you can pass as the first paramenter either ff or ogm to test the respective TPC-W implementation. After ensuring that everything is ok by manually running the previous steps at least once, scripting is most useful to speed up things.

#!/bin/bash

# CATALINA_HOME should point to your tomcat installation and is usually
# already set in the environment

# SERVER and CLIENT can be set here
TPCW=~/tpcw-benchmark/tpcw
SERVER=${TPCW}/server
CLIENT=${TPCW}/client

function stop_server() {
    ${CATALINA_HOME}/bin/shutdown.sh
    sleep 5;
    killall -9 java # careful!
    sleep 5
}

function reboot_server() {
    ${CATALINA_HOME}/bin/shutdown.sh
    sleep 15;
    killall -9 java # careful!
    sleep 5
    ${CATALINA_HOME}/bin/startup.sh
    sleep 3;
}

function build_client() {
    cd ${CLIENT}
    ant dist-client-only
    chmod 700 dist/rbe.sh
}

# pick a default backend if none is given
BACKEND_NAME=$1
if [ "${BACKEND_NAME}" = "" ]; then
    BACKEND_NAME="ispn"
    BACKEND_TO_USE="pt.ist.fenixframework.backend.infinispan.InfinispanCodeGenerator"
fi
case $BACKEND_NAME in 
    "mem")
        BACKEND_TO_USE=pt.ist.fenixframework.backend.mem.memCodeGenerator
        ;;
    "ogm")
        BACKEND_TO_USE=pt.ist.fenixframework.backend.ogm.OgmCodeGenerator
        ;;
    "ispn")
        BACKEND_TO_USE=pt.ist.fenixframework.backend.infinispan.InfinispanCodeGenerator
        ;;
esac

echo "USING BACKEND: ${BACKEND_NAME}: ${BACKEND_TO_USE}"

# ensure the client is build and ready
build_client

# start the web server
${CATALINA_HOME}/bin/startup.sh

NUM_EBS=10
# Mixes: 0=read-only; 1=browsing; 2=shopping; 3=ordering
MIXES="0 1 2"

# build and deploy the server app
cd ${SERVER}
mvn clean package tomcat6:deploy -DNUM_EBS=${NUM_EBS} -DNUM_ITEMS=1000

# run a few tests
cd ${CLIENT}/dist
for mix in ${MIXES}; do
    time wget -q -T 0 -O - "http://localhost:8080/tpcw/TPCW_populate?NUM_EBS=${NUM_EBS}"
    \rm -rf ObjectStore/ PutObjectStoreDirHere/
    ./rbe.sh -r workload-${BACKEND_NAME}-mix${mix} -i 120 -t ${mix} -b 1000 -n ${NUM_EBS} -tt 0 -u 60 -d 20 -w http://localhost:8080/tpcw/
    reboot_server
done
# undeploy the server app
cd ${SERVER}
mvn tomcat6:undeploy

stop_server

echo "ALL DONE. Check the results in *.m files located at ${CLIENT}/dist"

tpc-w's People

Contributors

pruivo avatar

Watchers

 avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.