Using AWS for Instructional Computing

Using AWS for Instructional Computing

Amazon Web Services

Amazon Web Services (AWS) provides a number of cloud-based services. In particular, instructors may be interested in using virtual machines and virtual computing clusters through Amazon's EC2 or MapReduce (EMR) services. SCF staff will provide individualized support to instructors wishing to use AWS in their classes.

Some potential instructional uses for AWS include:

  1. Working with cloud-based virtual machines
  2. Running parallelized computational tasks that scale to many cores using cloud-based virtual clusters
  3. Using Spark to carry out distributed computation, including MapReduce computations
  4. Using Amazon's MapReduce service for distributed computation

We note that SCF supports the use of virtual machines on student laptops and SCF servers, as well as provides access to parallel computing resources on our multi-core servers and our Linux multi-node clusters. So uses #1 and #2 above could be achieved either through AWS or through existing SCF resources. Doing it through AWS provides students with experience in cloud-based computing that may be useful in their careers. Doing it using SCF resources may be logistically easier.

Getting Started

Amazon's handling of credits for classes is currently in flux. The instructions below are out of date as the educational grants program does not formally exist any more. However, it may be possible to get a set of credits for a class via direct contact with Berkeley's campus AWS representative and follow steps 6-8 below. SCF can help with this -- please just email us.

To make use of AWS for your class, these are the steps you'll need to follow.

  1. Let us know of your plans and needs by emailing consult@stat.berkeley.edu.

  2. The cost of using AWS resources, in particular computing cycles with EC2 or EMR, can add up fairly quickly. To cover the cost of the class usage, please apply for an Amazon educational grant by going to http://aws.amazon.com/grants, clicking on the "Educators" tab and filling out the form. Amazon will generally provide a credit of $100 per student. Ideally, you would do this a couple weeks before you wish to use AWS in class as it can take a few days for Amazon to approve the grant.

  3. After 1-4 days, you'll receive an email from Amazon that should begin as follows. "Thank you for your AWS course grant application, your submission was successful. Would you like the credits set up for your AWS account and for you to manage access for students, or would your students be setting up their own AWS accounts separately?" Respond to the email and ask that credits be set up for your AWS account with access managed centrally. SCF will manage this access for your class.

  4. Wait another few days. You should get an email titled "AWS In Education Grant - Congratulations!" which will contain a credit code resembling "PC1IJ749GVDDG09".

  5. Send email to consult@stat.berkeley.edu containing:

    1. the credit code
    2. a plain text file containing the @berkeley.edu email addresses of your students (obtained from BearFacts)
    3. a separate plain text file containing the @berkeley.edu email addresses of any instructors (including GSIs) who should have administrative access

     

  6. SCF staff will then apply the credit to the overall SCF AWS account and will make accounts for the students and instructors/GSIs. Each individual will receive an automated email containing a URL to login to, their user name, and a one-time password to use the first time they login. Once the individual logs in, they will be prompted to create a new password to access the AWS website and will be able to download the AWS credentials (an AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY) that can be used to start EC2/EMR virtual machines/clusters.

  7. Instructors will be able to monitor activity in EC2 or EMR by logging into the AWS Management Console. Go to http://aws.amazon.com and click on "AWS Management Console" under the "My Account" pulldown. Login using your username and your password. Then click on the orange cube in the upper left corner and select the service you'd like to access (e.g., EC2).

  8. Users can start up virtual machines/clusters in a variety of ways: by using the AWS web interface, using scripts provided by Spark, using the StarCluster command line tools, the Python boto package and many others.

SCF staff can provide technical assistance with actually using AWS in your class as well as general feedback on how you might use AWS for instruction. Please email consult@stat.berkeley.edu with questions.