We have worked to develop a process to simplify cloud adoption by Rensselaer faculty and researchers. This includes assisting you with getting your groups account setup, the initial configuration and providing recommendations on technology and cost saving.
Your group is responsible to pay for the usage charges incurred. We have setup a process where DotCIO handles billing from the vendor on your behalf and settles the payments via internal funds transfer.
- Azure provides several technical and process benefits that make it our provider of choice for cloud workloads.
- AWS is available to be used for specific use cases that cannot be met using the Azure platform.
- GCP is not currently available. If you have a use case that requires this provider let us know.
- The cloud is about service delivery, not individual machines
- Running persistent (permanent) virtual machines in the cloud is expensive
- Data should not live inside virtual servers, they are meant to be disposable
- Running containerized, server less, or cloud native, services is inexpensive and scalable
- You should be using software and config versioning in git with your cloud project
- The cloud does not mean public. Services can be run as secure and private as on-premise
- Cloud costs are variable
- Shared responsibility model
How to start
If you are interested in using one of these cloud providers as part of course delivery, research, or other RPI projects open an ITSSC ticket with the subject "new cloud project".
We will work with you to understand your goals, setup your subscription, estimate costs and create a budget to track your spending. We will need the FOPAL and business manager approval before beginning. We are also asking that department heads be CC'd on the request.
Recommended general use cases
The cloud provides for infinite options but there are a couple that we have found currently do and do not work well.
- Providing resources to students during class or lab. Such as web servers, computers, etc.
- Containers. Any application that has been built to run as a container can be run, and scale very effectively in the cloud.
- Archival data storage. The cloud is very cost effective for long term storage of infrequently accessed data.
Not so good
- HPC and Large GPU systems have not proven to be cost effective in the cloud. Our estimation is that they cost 8x our on premise resources for long running needs. They do make sense for certain workloads, particularly if the need is short or if a proprietary technology or toolchain is only available in a cloud native format. We continue to evaluate this and would be willing to help evaluate your specific use case.
Purpose build use cases
We are developing a set of purpose build environments to be easily reproducible for individual classes or projects. These environments are fully integrated with RCS authentication, course enrollment data feeds, and RPI's GitHub version control environment.
Student Virtual Machines
Provides individual virtual machines to each student in a course. Also supports grouping students into teams with a shared machine.
- Students have full access to configure the OS inside the VM
- The environment can be pre configured by the instructor to facilitate course tasks
- Students can turn the machines on and off as needed and rebuild from original state
- Machines can be configured to be accessible from the internet
- Instructors have the ability to limit which machine instance sizes a student can configure
- Instructors can power on machines to facilitate grading or other access without student intervention
Last Reviewed: 17-Oct-2022