As many of you may already know, VMware just released the VMware Appliance for Folding @ Home Fling last week and you can check out this blog post A Force for Good: VMware Appliance for Folding @ Home by Amanda Blevins for all the details. As everyone is wrapped up in their own personal quarantine, we want to show you how we in the Hands-On Labs are pushing the limits of our hardware to help the Folding@Home program understand protein folding, the diseases that result from protein misfolding and aggregation, and novel computational ways to develop new drugs in general.
The team running the Hands-On Labs have been experts in infrastructure automation for years, so we knew we could reserve a chunk of our capacity to help push this project easily! As we host various events such as vForums and VMworld: our team must have capacity at the ready to spin up all of the labs you know and love. After checking with our teams and forecasted capacity needs we were able to gather 3 different clusters totaling nearly 3500 logical cores which equated to nearly 36 THz worth of compute to deploy. That’s like having 18000 laptops crunching away!
As you’ll notice right now nothing is really being used. That’s compliments of the way VMware software allows companies to easily spin up capacity on demand. The Hands-On Labs run on a combination of VMware Cloud Foundation, vCloud Director and VMware Learning Platform to deliver the labs. This great combination of software gives end users simple controls when it comes to scaling workloads up and down. To show you how easy it is, I’ve documented the steps I took in a day to spin up one of our environments.
- Download the VMware Appliance for Folding @ Home from my browser.
- Upload the appliance to vCloud Director using the “Add vApp From OVF” button.
- Customize the Appliance for Folding@Home template to my liking by setting the OVF Properties.
Note I made sure the team was set to 52737 Team VMware!
- Add the template to my vCloud Director Folding catalog.
- Over on the VMware Learning Platform – I then Sync my vApp Templates with vCloud Director.
- Finally – Create a deployment pool in VMware Learning Platform.
- Last – but certainly not least: Watch your templates populate in VMware Learning Platform and vCloud Director!
Many of you may notice that we’re only running giant VMs for our contribution for Folding@Home. This is due to the fact that we are looking to maximize our ability to fold by minimizing context switching across VMs. It also allows us to contribute to the BigAdv type of workloads that get far fewer opportunities due to a lack of available CPU cores (16 or more!!!).
That’s it! The HOL now has reserved a massive amount of CPU capacity specifically for Folding@Home. When our next big event comes up, we can pare down the workload VMs that are sitting idle to give capacity for events, and then spin it right back up as soon as we no longer need the capacity.