How a large hedge fund uses Nomad to arbitrage the cloud
Dec 05, 2018
Using Nomad paired with Terraform for uniform multi-cloud provisioning, this firm chooses the cheapest cloud vendor and instance type each day to save on costs.
VP Worldwide Solutions Engineering, HashiCorp, HashiCorp
There's a large hedge fund that has looked at Nomad as a tool that allows them to achieve the scale that they needed to achieve. As you can imagine, a hedge fund has a need for efficiency as well as cost optimization so that they can run all of the different workloads that they need to be able to do their financial modeling.
This hedge fund looked at HashiCorp and the tooling that we provided and said, "Can we leverage Terraform to be able to essentially arbitrage the cloud?" So, if you look at GCP and Azure and AWS and say, "What is cheapest that day? Can we provision 200,000 cores in that cloud to be able to run workloads on?" So, they did. And they built out those Terraform templates and every day, 20 times a day, they look at whatever the cheapest price is and will provision 200,000 cores.
The next challenge that you face is, how do I actually leverage those 200,000 cores? I don't want to leave them up for a day, or two days, or even multiple hours because it becomes very expensive. But what they are able to do with Terraform by spinning it up really quickly and tearing it down really quickly is save costs.
By leveraging Nomad, what they can do with the batch scheduling is be able to take those 200,000 cores and run batch workloads and do those very efficiently where they're doing 5,000 containers a second to be able to process their financial modeling for that day. And then they tear everything down.
You do that 20 times a day and you, as a hedge fund, are trading off of much richer and much more recent financial modeling data than any other company because you're leveraging the compute available to you by standardizing on HashiCorp tools.