Penn Computing
Computing Menu Computing A-Z
Computing Home Information Systems & Computing Penn

Amazon Web Services

 

aws Amazon Web Services (AWS) is an 'Infrastructure as a Service' (IaaS) provider. AWS' cloud based computing offers dynamically scalable computing, storage, and data access “on demand” over the Internet, with “pay as you go” pricing for the hardware and software that is delivered. ISC has established a master services agreement with Amazon for use of it's various services. AWS includes over two dozen cloud-related services, including their popular elastic computing (EC2) and storage (S3) capabilities. Broadly speaking, their cloud services can be grouped into 3 categories, infrastructure services, platform services and supporting services. A few notable highly used services are described below.

 

EC2

Elastic Compute Cloud is Amazon's name for it's service that provides virtual machines on demand. As all AWS services, this service can be used through a web portal interface, a command line, a software development kit (SDK) as well as an application programming interface (API). The EC2 service allows users to create, start, use, stop and destroy virtual machines in any of Amazon's data centers. AWS uses a sophisticated architecture to provide highly available and redundant locations for the virtual machines. Within EC2, AWS provides a distinct set (although large) of specific virtual machines configurations. Many of these configurations are tailored toward specific use cases, such as cluster computing or database servers.

EBS

Elastic Block Store (EBS) provides block level storage volumes for use with Amazon EC2 instances. So, the EBS storage volumes you create (in sizes from 1 Gb to 1 Tb) can be mounted as devices by Amazon EC2 instances. Multiple volumes can be mounted to the same instance. To the EC2 virtual machines, the volumes look like raw storage devices, much like the hard drive in a conventional computer. The volumes exist and persist independently of the virtual machines in the EC2 service.

S3

Simple Storage Services provides a structured storage service mostly targeted at web applications. It allows for storage of objects between 1 byte and 5 terabytes in size. The number of objects you can store is unlimited. Each object exists in a 'bucket' and can be managed (create, read, update, delete) through a unique, developer assigned key. The objects in the S3 services can be stored in multiple regions, depending on the requirements (such as regulatory, latency or cost considerations) of the application.

Cost

AWS charges for any use of their services, usually by the hour or any part of an hour that a service is used. Amazon also charges for the bandwidth used for data transfer between their services and the Internet As one example, this means that if you use a medium sized Linux VM, which costs $0.160 per usage hour, unless you shut the machine down whenever it is not used, the cost of running that VM will be $0.16 * 24 * 365 = $1401.60 per year. As another example AWS S3 storage costs $0.125 per Gb per month for the first Tb. If you were to store a Tb of data at AWS, this would cost about $0.125 * 1024 * 12 = $1536 per year. AWS provides many different price tiers and ways to reduce this cost by reserving instances or choosing less redundant storage, but be advised that managing the usage of AWS resources on an ongoing basis is very important to the financial bottom line. AWS is best suited for systems have a 'spiky' usage profile.

Master Services Agreement

The services that AWS provides to the University are part of a larger effort being led by Information Systems and Computing (ISC) to actively develop relationships and advantageous terms with leading cloud-computing providers, as well as to develop and make available supporting internal Penn services to ensure optimum advantages for the Penn community. This and future enterprise agreements provide protections and advantages not available in individual “click-through” agreements. A University wide umbrella agreement provides:
  • Security and Privacy: Amazon has made a binding commitment to strong privacy, security and compliance measures that support Penn’s policies and recommendations.
  • Termination and Suspension: Accounts can no longer be unilaterally suspended or terminated and must meet strict definitions for cause. Suspensions that do occur will be limited in scope.
  • Data Recovery: In the event account termination is necessary for any reason, all Penn users are guaranteed 30 days to retrieve data.
  • Billing, Reporting and Auditing: Centralized billing, account management and reporting are available.
This master services agreement addresses the risk of individual usage of AWS through non-negotiable “click-through” agreements, which require an individual to agree to Amazon terms and conditions. The non-Penn specific standard terms made AWS unsatisfactory for most Penn uses from both a legal and operational perspective.

AWS Training

As part of the overall master agreement, Penn has negotiated for Amazon to provide resources to assist in on-campus orientations and training for the full set of AWS features and services. This will include an orientation and introductory training to all AWS services for any interested Penn stakeholder, as well as more formal, detailed sessions for IT personnel and any interested faculty, staff or students. Training opportunities will be scheduled and publicized.

Additional Information

Individuals and organizations working with Penn data are urged to use AWS only when protected under the master service agreement.

If you are a member of the following schools please contact your school's master account holder:

For others wishing to take advantage of the Univsersity's master service agreement please contact your local IT Organization or Michel van der List

Please see https://aws.amazon.com/ for more details on the full scope of AWS services.