Tutorial: Benchmarking Raddit

Raddit is a clone of Digg for use in Berkeley RAD Lab cloud computing research. This tutorial will help you use Rain to benchmark multiple servers running the Raddit web application.

Requirements

• A “web server” with Rails (v2.3.4), Thin (a lightweight Rails server), and Nginx (a load balancer).
• A “database server” with MySQL.
• A “workload server” with Java.

Deploying Raddit

1. SSH into the “application server”.
2. Checkout Raddit from the repository.
   • Newer versions of Git support the --recursive argument.
     1. Use: git clone --recursive git://github.com/yungsters/raddit.git
   • Older versions of Git do not support the --recursive argument.
     1. First, clone the repository: git clone git://github.com/yungsters/raddit.git
     2. Change to the cloned directory: cd raddit
     3. Checkout the submodules: git submodule update --init
3. If you haven’t already, change to the Radddit directory: cd raddit
4. Use a text editor to update the production database configuration in config/database.yml.
5. Create the production database: rake db:create RAILS_ENV=production
6. Load the database schema: rake db:schema:load RAILS_ENV=production
7. Start the Thin servers: thin start -e production -s <number of Rails servers>
8. Configure Nginx to point to localhost on the Rails ports.

Running the Benchmark

1. git clone git://github.com/yungsters/rain-workload-toolkit.git
2. cd rain-workload-toolkit
3. ant package
4. ant package-raddit
5. java -Xmx1g -Xms256m -cp "rain.jar:workloads/raddit.jar" radlab.rain.Benchmark <path to Rain JSON configuration file>

See the instructions on the Getting Started page for more information.