CSC/ECE 517 Fall 2013/ch1 1w09 hs

Code Deployment and Release Management with Rails

Code Deployment or Software Deployment and Release management with Ruby on Rails can be done using various tools such as Capistrano, Heroku, AppFog, OpenShift, AWS, Cloud 66, Vlad, etc. These automated tools are frameworks used to write and execute commands in parallel remote machines and thus ease the process of deployment and management of various applications.

Introduction

The Deploying process involves different stages like development, testing, release management and sign-off by project teams.

The following diagram illustrates the deployment process of an application.

<ref>http://technet.microsoft.com/en-us/library/cc465178.aspx</ref>

The deployment flow has three stages: development and testing, release management and stakeholders and project teams.

After the final release of the code in development and testing stage, the master plan is given for deployment of components and sites based on the signoff sheets given by the project teams, stakeholders and customers. Then, the stabilization of the deployment is calculated based on some metrics. Once the deployment is stabilized, it is reviewed and the process of deployment is completed. the final signoff form is given by the stakeholders.

Code Deployment

Code deployment is a procedure where the application is transferred onto a production server, so that it is made available for all the other users. It is the part of release management process which involves several activities. These activities are inter-dependent and need to be completed in a specific order for successful code deployment.

Code Deployment Activities

The various deployment activities include <ref>http://en.wikipedia.org/wiki/Software_deployment</ref>

• Release
• Install and activate
• Deactivate
• Adapt
• Update
• Built-in
• Version tracking
• Uninstall and retire

The general guidelines for the deployment process remain the same but there are variations in the specific requirements with each software system.

Release Management

<ref>http://searchitchannel.techtarget.com/definition/release-management</ref> Release management is a software process which takes care of the development, testing, deployment and support stages of software releases. The start of release management happens in the development phase, where it requests for new features or changes in the existing application. When the request is accepted, a new release is planned and designed. This design is built and then enters the testing phase, where the release is tested until it is accepted. After the release is accepted as a candidate, it enters the deployment phase where it is implemented. Once deployed, the release is made available and sent to support phase for collection of bug reports. Once the support phase is done, the bugs and other issues lead to requests for changes and the cycle is then repeated.

Automating Code Deployment and Release Management

<ref>http://www.urbancode.com/html/solutions/deployment-automation.html</ref> Automation of a process is required when the manual process is more error prone. Also, for large real world applications, the manual process is very complex and time consuming. Incomplete documentation hampers a manual process further. Deployment also has to be done in many environments and involves varied steps to be followed in a specific order. Manual process needs repetition of same procedure to each and every environment which leads to many errors and the process becomes costly. In web applications each new release has to be deployed to and tested in development, testing and production environments.The number of software releases in web applications are more and hence the need to automate release management is also greater.

Automation of deployment delivers applications faster and with fewer errors. The defined software deployment process can be applied across all environments consistently every time. With various automated application deployment platforms now readily available, manual deployments have become a thing of the past.

Deployment Automation tools

Some of the automated tools that are widely used for the process of deployment with Ruby on Rails are

• Capistrano
• Heroku
• Vlad
• AppFog
• AWS Elastic Beanstalk
• OpenShift
• Cloud 66

Capistrano

Capistrano Logo

Capistrano is a Ruby tool which helps deploy an application to the server. Capistrano allows deploying to multiple machines at a time. It has many advanced options which help to deploy different kinds of applications. It also supports the scripting and execution of tasks.

Capistrano can be used to:

• Deploy applications onto many machines simultaneously.
• Automate the audits
• Automate the common tasks of a team.

Capistrano can be made a part of a larger software by integrating it with another Ruby software.

<ref>http://www.capistranorb.com/documentation/getting-started/installation/</ref>   General Usage

The following commands will clone Capistrano, build the gem and install it locally.

$ gem install capistrano --pre --trust-policy HighSecurity

Or grab the bleeding edge head from:

$ git clone -b v3 https://github.com/capistrano/capistrano.git
$ cd capistrano
$ gem build *.gemspec
$ gem install *.gem

Signed Rubygems

As Capistrano is a signed gem, you should always be careful to use the --trust-policy flag when installing Gems, or since Bundler 1.3 you should use the same flag:

$ gem install capistrano --pre --trust-policy HighSecurity
$ bundle install --trust-policy HighSecurity

If you get a message that looks like:

ERROR:  While executing gem ... (Gem::Security::Exception)
unsigned gems are not allowed by the High Security policy

Then please complain to your Gem author, and have them start signing their Gems.

Usage in a Rails project

Add the following lines to the Gemfile to the :development group ideally.

group :development do
  gem 'capistrano-rails', '~> 0.0.7'
end

Heroku

Heroku Logo

Heroku is a cloud Platform as a Service (PaaS) which supports several programming languages. It helps manage environment-specific configurations separately from the source code, which provides greater security. Heroku is not just a deployment environment but it also impacts the development process. There are several principles of the development of application on Heroku.
• Applications and Codebases : Most of the development of the application is done on application’s codebase which is stored in a Version Control System (VCS)
• Dependencies : All the application dependencies such as plugins should be declared explicitly
• Configuration : Configuration of an application is all that varies in different deploys

Heroku CLI

Heroku CLI is managing Heroku applications over command line.

Once Heroku CLI is installed, heroku commands can be accessed from the command shell. User has to Log in using the email address and password used when creating the Heroku account:

<ref>https://github.com/heroku/heroku</ref>

$ heroku login
Enter your Heroku credentials.
Email: adam@example.com
Password:
Could not find an existing public key.
Would you like to generate one? [Yn]
Generating new SSH public key.
Uploading SSH public key /Users/adam/.ssh/id_rsa.pub

Deploying Applications

Heroku uses git primarily to deploy the applications. When an application is created on Heroku, it associates the new git which is remote with the local git repository.

As a result, deploying code is just the familiar git push, but to the heroku remote instead:

<ref>https://devcenter.heroku.com/articles/how-heroku-works#deploying-applications</ref>

$ git push heroku master

Vlad

Vlad Logo

Vlad is another automated deployment tool which is simple to use. It integrates with rake and uses standard tools like SSH, rsync etc.

The main features of Vlad include

• It uses very few dependencies which are simple.
• It can execute commands on more than one server.
• It syncs files to one or more servers.
• It matches local and remote tasks.
• It runs tests very fast.

<ref>http://docs.seattlerb.org/vlad/index.html</ref> Running Vlad

• Using rake

task :shazam! do
  Rake::Task[:action1].invoke
  Rake::Task[:action2].invoke
end

• Using SSH

Host example.com
    User fluffy_bunny

AppFog

AppFog Logo

AppFog is a PaaS built on Cloud Foundry which is another PaaS platform for deploying applications.

Features of AppFog are

• It launches and runs fast.
• It does not require configuring servers or installing frameworks.
• It is compatible with code management systems like git, svn etc.

Installation

‘af‘ command line tool is written in Ruby for installation

Getting Started

target [url]                                       Reports current target or sets a new target
login [email] [--email, --password]                Login
info                                               System and account information

AWS Elastic Beanstalk

AWS Logo

Amazon Wen Services(AWS) offers Infrastructure as a Service (IaaS) and provides a complete set of resources for the development of a web application in minutes. AWS Elastic Beanstalk is a deployment tool of the AWS cloud.

With AWS Elastic Beanstalk, applications can be quickly deployed and managed in the AWS cloud without worrying about the infrastructure that runs those applications. AWS Elastic Beanstalk reduces management complexity without restricting choice or control. Once the application is uploaded AWS Elastic Beanstalk automatically handles the details of capacity provisioning, load balancing, scaling, and application health monitoring. AWS Elastic Beanstalk uses highly reliable and scalable services.

AWS CLI

AWS Elastic Beanstalk uses a command line tool.

Run the sample using a command line interface. For example:

eb status

The above command gives the status of the application in the deployment process.

OpenShift

OpenShift Logo

Another free and open source PaaS for deploying applications is OpenShift created by [RedHat].

Considerations

• Databases

To use the application outside the OpenShift environment one needs to change the variables as the application is configured to use database in production mode in OpenShift.

• Assets

The assets are precompiled everytime application is pushed to OpenShift.

• Security

Security related variables are to be unique across the application.

<ref>https://github.com/openshift/rails-example</ref> Installation

1. Create an account at http://openshift.redhat.com/

2. Create a rails application

rhc app create -a railsapp -t ruby-1.9

3. Add mysql support to your application

rhc cartridge add -a railsapp -c mysql-5.1

4. Add this upstream Rails quickstart repository

cd railsapp
git remote add upstream -m master git://github.com/openshift/rails-example.git
git pull -s recursive -X theirs upstream master

5. Push your new code

git push

Cloud 66

Cloud66 Logo

Cloud 66 combines the convenience of Platform-as-a-Service (PaaS) and the flexibility and control of Infrastructure-as-a-Service (IaaS). This platform allows easy deployment of application, so there is no overhead of configuring or monitoring the servers.

Cloud 66 Command line Toolbelt

Toolbelt is an open source released to deploy the application.

<ref>http://blog.cloud66.com/</ref> Installation

Installation of toolbelt uses the following command

Gem install c66

This automatically searches the newest version and installs it.

Comparison between Capistrano/Vlad/Heroku/AppFog/OpenShift/Cloud66/AWS Beanstalk

With so many automated tools available for Code Deployment and Release Management, the decision on which to use for a particular Ruby on rails application depends on various parameters to be considered. Some comparative features of the different tools are as follows :

Capistrano	Vlad	Heroku	AppFog	OpenShift	Cloud66	AWS Beanstalk
It is a framework that allows one to execute commands on a remote server using tools like ssh.	Vlad deployment is similar to Capistrano, but less complex.	Heroku’s offers a plan with first dyno free i.e. 512 MB that can be paired with Postgres starter DB to start an application with 0 initial cost	AppFog’s initial plan offers 2GB of RAM free which is enough for small to medium traffic applications. By managing options provided for DB, queue server, application server, etc all part of the intial package, this is an excellent option to try out an application	OpenShift is an open source platform and its free plan for starters has 3 gears for free wherein each gear consists of some RAM and disk space.	Free plan provides one server for deployment.	Provides an initial free tier. Without free tier no additional cost for Beanstalk. One only needs to pay for the AWS resources used.
Memory increase would require additional servers/unused space in existing ones.	Memory increase depends on additional servers available/unused space in existing ones.	Memory increased in terms of dyno each of size 512 MB.	Can configure any size for memory allocation	-	-	Can automatically up/down scale the application by adjusting the default auto scaling settings.
While this provides greater flexibility as user is in control of the servers, it also causes additional overhead in terms of user needing to work both on Rails application and web server deployment.	It can be integrated with Rake without any hindrances.	Can dynamically scale up and down an application and pay as per hourly usages. Each additional dyno costs $35/mo	Scaling takes up some time and need to pay for a month at the minimum. Scaling up to 4GB increases cost to $100/mo	-	Scaling using Cloud66 is relatively easy and hassle free.	Scaling is fast and easy.
For sensitive applications where handing over the control to a third party for deployment is not feasible, Capistrano is the way to go.	It also provides support for tools like rsync and ssh.	Lacks support for cross region deployment i.e. can’t select location of server where application gets deployed	Supports cross region deployment i.e. if users of application primarily in Asia, can choose a server in Asia to deploy application, thus making the site faster.	-	-	Provides complete flexibility and control of the resources to the user including the chance to take over control of some or all.
Provides support for DB related to rails applications	Provides support for DB related to rails applications	It features serious production DB like RDBMS, Postgres, etc. and many Addons such as MongoDB, ClearDB, etc.	Features free shared databases such as MongoDB, MySQL, Postgres but for other production DB’s need to use Addons	Databases featured are MySQL, MongoDB and Postgres	-	Supports DB like Amazon SimpleDB, Amazon RDS, Amazon DynamoDB.
Capistrano provides commands to be executed for automated deployment. It supports git.	It is much easier to deploy using Vlad as compared to Capistrano.	Deployment as easy as ‘git push’	Deployment does not support git. It uses the af tool for deployment.	-	Since Cloud66 provides the platform in terms of Application Stack Management as a Service, it uses the infrastructure of AWS, Virtual Private Server, etc. for actual deployment. But the user doesn’t have to worry about configuration or monitoring of the servers. Hence provides easy deployment.	Deployment options include Git, AWS management console, etc.
As it has been in use for a long time there is a huge community support available for Capistrano.	-	Heroku also provides a huge community support given the fact that there are a lot of Heroku users.	Community is limited.	RedHat being a reputed provider for open source solutions, the OpenShift community is gradually increasing and receiving support from a considerable community of users.	Limited community support currently	Has community forums and documentation.

Conclusion

Comparison and contrast between various deployment automated tools such as Capistrano, Heroku, Vlad, AppFog, OpenShift, Cloud66, AWS Beanstalk shows that each has its own pros and cons. Various considerations about the application such as scalability requirement, security concerns, cost, availability of own resources such as servers, etc will help determine which tool is is the most suitable for that application.

References

<references/>

See Also

• Rails Solutions: Ruby on Rails Made Easy by Justin Williams
• http://guides.beanstalkapp.com/deployments/best-practices.html
• http://www.urbancode.com/anthill3-help-3.7/html/SettingUpDeploymentProcess.html
• http://livedocs.adobe.com/coldfusion/8/htmldocs/help.html?content=UseFMSGateway_3.html
• http://pic.dhe.ibm.com/infocenter/tivihelp/v26r1/index.jsp?topic=%2Fcom.ibm.dtm.doc%2Fdeploysoftware%2Fc_deploysw_process.html