How you can help battle COVID-19 with idle servers

[email protected] is a research consortium that uses open-source, distributed computing to study and develop treatments for diseases like cancer, Alzheimer’s, Huntington’s and Parkinson’s—and now COVID-19. The organization’s innovative research model uses idle and underutilized CPU, donated by individuals and businesses, to run simulations of protein dynamics to help researchers better understand the biology of diseases. This is critical as proteins having many moving parts that cannot be studied through traditional research that only provides a single, snapshot-in-time view of the protein. By running and studying millions of simulations, researchers can study proteins in action to reveal treatment opportunities. 

Since taking on COVID-19, [email protected] has increased its number of contributors by 1,200% for 2.6 EXAFLOPS (or 2.6 quintillion calculations per second.) While this is a tremendous amount of capacity, more contributors means more acceleration to findings. 

Flexential became involved with the [email protected] movement in late March as office closures and stay at home orders were going into effect. Kourosh Mobl, a Senior Engineer, and Kevin Hahn, an Enterprise Data Architect, approached the organization with the ideas of donating idle CPU/GPU cycles from the labs of various organizational teams, including security engineering, platform design and professional services. 

We are also leveraging our NetApp ONTAP AI partnership with Trace 3 to run high-density work units on spare cycles of our NVIDIA DGX cluster in our Portland - Hillsboro 2 “Gen 4” data center. This cluster comprises four very powerful DGX-1 servers, each containing eight Tesla V100 GPUs, and uses a Kubernetes cluster to run the [email protected] software, setup and configured by Flexential professional services.

GPUs like those from Nvidia are suited for these types of computational heavy use cases. It’s why they are the favored platform for machine learning and AI applications.  

To put the power of the DGX into perspective, over the past 30 days the cluster has completed 4,300 work units to rank Flexential consistently in the top 75 producers out of 250,000 contributors. Flexential’s infrastructure is designed to support these high-density workloads, while its ONTAP AI partnership simplifies AI deployments by eliminating design complexity. To read more about our ONTAP AI partnership and the DGX clusters that power it click here. 

So let’s help [email protected] reach its goal of one million folders. Getting involved is easy and will not disrupt your own work. Simply download the [email protected] client to your computer and join one of the thousands of already-established teams that are running simulation work units. 

VMware recently introduced a [email protected] appliance deployed via vCenter to allow corporations that manage their own server infrastructure to participate. While this may not be viable for every business, it is worth asking for your leadership’s support. 

Together we can work to find a cure for COVID-19, one CPU or GPU at a time.