Carbon aware cloud computing for our high performance computing workloads

Ortec Finance’s cloud software results in many high-performance computing workloads. This results in high energy consumption.

From our desire to limit the impact of our high-performance computing workloads on our planet, and finding out there currently is no proper solution for this on the market, we decided to build one ourselves.

Ortec Finance started a project called Sailfish: we chose a solution of a combination of managed operators and some do-it-yourself components. For the managed ones we chose for Red Hat, a partner of us.

The project led to the following current situation: At start of a large simulation we compare our own with other region’s grid intensity, e.g. defined by CO2e/kwh. Is there a region with a better grid intensity? Then we move the workload to that specific region. The estimated energy consumption together with grid intensity determines the carbon emission. With this optimization tooling we can have our high-performance computing workloads executed in areas resulting in the lowest carbon emission.

Lisette van Leeuwen explains how this works in this video, shot while presenting about our Sailfish project at OpenShift Commons Gathering at Red Hat Summit 2024, Denver.

Lisette: “Carbon aware scheduling is important to us, because we as an IT company have a responsibility to perform our simulations in regions where datacenters are powered by renewable energy.”

Interested in learning more about carbon aware cloud computing?

Contact Lisette van Leeuwen or Joris Cramwinckel