Researcher Virtual Machines (VMs) provide a dedicated environment for researchers to either prototype code that will eventually run on the VACC or run production code/applications that cannot run on the VACC. These machines provide the flexibility for researchers to scale to meet their needs. The VMs are also supported by a dedicated team within the Systems Architecture and Administration group.
What are they?
Research VMs are virtual computers that run on hardware in the UVM datacenter. They run a full operating system and allow for remote access through a variety of tools. They have a dedicated amount of resources (CPU, RAM, storage) that can scale up to meet researcher needs over time. Unlike the VACC, the resources dedicated to your VM are only available to you and there is no queue for work.
Who is eligible?
Faculty members can request a VM for their personal use or use within their research group. In order to keep management simple, graduate students and post-doctoral candidates cannot request machines directly, they must work with their advisor/PI.
Research VM Benefits
- Highly available
- Scalable
- Backups available upon request
- Supported/managed by systems administrators
- Customizable
These VMs reside on a highly available, dedicated cluster of hosts using VMware technologies. They are backed up and can be configured with different levels of reporting/monitoring to ensure uptime and reliability. Core operating system maintenance is managed by ETS, so there is no need for researchers to worry about patching critical security updates, it's all taken care of. Removing this maitenance responsibility means researchers and their students can focus on research instead of system administration.
Your VMs can be customized to fit your exact research needs with a large variety of applications and tools. These VMs are designed to be a replacement to dedicated physical research workstations that can take up space and age out quickly. VMs also scale much easier than a physical workstation. For example, increasing memory or CPU count on a VM can be done on the fly without a reboot. Upgrading storage can take about 10-20 minutes without the need for re-imaging or any potential data loss.
Pricing
The VM platform is very flexible and specs can be customized to fit exact researcher needs. The configurations below are only examples and are not the only options available.
| Free Tier | Example #1 | Example #2 |
|---|---|---|
12 CPU Cores 48 GB RAM 100 GB Primary Storage | 16 CPU Cores 64 GB RAM 100 GB Primary Storage 4 TB Netfiles2 | 32 CPU Cores 128 GB RAM 400 GB Primary Storage 10 TB Netfiles2 |
| Free1 | ~$2,990 | ~$6,525 |
This pricing is a one-time cost that is good for 5 years of use. Researchers may choose to not use their free VM and will instead get a $2,000 discount on their first VM.
1 Research faculty or full-time research staff are eligible for one free VM
2 This is in addition to the 10 TB of Netfiles all faculty get for free, learn more about free Netfiles storage here: https://www.uvm.edu/it/kb/article/research-storage/
VM Host Specs
5 x Dell PowerEdge R7525 hosts with:
2 x AMD EPYC 75F3 32-Core (2.95GHz) processor
1 TB RAM
25 Gbps networking
NVIDIA L40 48GB GPUs (in three of the hosts)
Currently supported operating systems are:
RedHat 8.x, 9.x
Windows Server 2019
Example Software Available:
Python
PyTorch
TensorFlow
Matlab
ifort (Intel-based Fortran)