Research Computing Expertise

Computing Services

This category consists of systems that will allow you to run data and analytics workloads. Below you will find a combination of both Linux and Windows-friendly systems to choose from.

Armis - This is a Linux-based high performance computing (HPC) option specifically designed to support users who are processing sensitive data. It operates very similarly to Great Lakes, but the additional security controls allow sensitive data (such as PHI) to be processed. Armis is equipped with GPUs for machine learning workloads.

Great Lakes - The University’s primary Linux HPC system, Great Lakes is brand new for fall of 2019. It replaces the previous HPC system, Flux, with brand new hardware upgraded for maximum performance. Users can run single-node jobs, or scale up to many physical nodes if they use software that can scale across many physical servers. Note that Great Lakes cannot be used for HIPAA data (use Armis instead). Great Lakes is equipped with GPUs for machine learning workloads.

Yottabyte - This system is what is known as a private cloud. Users can request both Linux and Windows virtual machines which can be remotely accessed from anywhere with an internet connection. Virtual machines can be scaled up and down to meet your computing needs, and they can also be configured to handle sensitive data. This is a more user-friendly option for users who are not comfortable with command line systems such as Great Lakes and Armis. However, note that Great Lakes does have graphical interfaces available via web browser. Yottabyte is not equipped with GPUs at this time.

Public Cloud

The University has service agreements with Amazon Web Services, Google Cloud, and Microsoft Azure. Each of these services offer their own benefits, and public cloud services provide a wide range of features that you won’t often find available on campus.

Please keep in mind that all three of these services are currently not certified by UM to handle sensitive data. The University is currently working on agreements to allow these services to house sensitive University data.

Feel free to reach out to us to find out if the public cloud might be the right option for you given your computing needs.

Storage Services

These services are specifically designed to store data. Researchers will often connect these storage services to the compute services listed above.

For those of you with lots of data, here are some great options.

Are you an employee on the Medical campus?

Shared3 - HITS offers a new storage service called Shared3. It can be configured to support sensitive data such as PHI. The service can also support users with multiple terabytes of space needs.

Are you an employee on the University campus, or planning to use any of the central campus computing services (Armis, Great Lakes, Yottabyte)?

Data Den - Managed by Advanced Research Computing (ARC), Data Den is an archive service meant for storing files for very long periods of time at very low cost. If you have files that you need to keep but do not plan to use in the near future, this is a good option for you. This service is provisioned in increments of 1 terabyte (TB).

Locker - Managed by ARC, Locker is a cost-optimized storage system designed for users with very large storage needs. This service is provisioned in increments of 1 TB with a 10 TB minimum purchase. On the cost vs performance spectrum, Locker sits in the middle between Data Den and Turbo, with Turbo being the highest performance and highest cost.

MiStorage - Managed by Information and Technology Services, choose this service if you need a general-purpose storage option and have less than 1 TB of data. MiStorage offers a moderate level of performance for most applications, and is great if you need an option that you plan to connect directly to your local workstation.

Turbo - Managed by ARC, this storage system is built to be a high-performance storage system. Choose this service if you either a) plan to use any of ARC’s other computing services for large data analysis, and/or b) you have more than 1 terabyte of data that needs to be accessed quickly and frequently.