No More Stalled Workflows! AWS HealthOmics Adds Automatic Resource Fallback

Hi, I'm Shii-chan! Today I found a nice update for the bioinformatics crowd!

AWS What's New aws.amazon.com

What was announced?

According to AWS What's New, AWS HealthOmics has added a "resource fallback" directive for WDL (Workflow Description Language) workflows. For each task, you can now define an ordered list of preferred accelerator types, with an option to fall back to CPU instances as a last resort. AWS HealthOmics is a HIPAA-eligible, fully managed bioinformatics workflow service for healthcare and life sciences customers.

The story so far

Until now, if the accelerator specified for a task wasn't available, the workflow would simply stall there. Researchers and bioinformaticians had to diagnose the cause, pick a different resource, and resubmit the run manually. Having a production workflow interrupted every time an accelerator was in short supply was a real pain point.

What changes

With resource fallback configured, HealthOmics automatically moves through your list of alternatives when your preferred accelerator isn't available, no resubmission required. That frees researchers from diagnosing and manually resubmitting runs every time there's an accelerator shortage, and keeps production pipelines running.

Dive Deep

There are a few knobs you can tune when setting up resource fallback:

• Each accelerator profile has its own configurable timeout
• A shorter timeout lets you move through the fallback list quickly
• A longer timeout increases the odds of reserving your preferred accelerator
• Adding a CPU profile as the final fallback further increases the chance of reserving an instance

Resource fallback is available now for WDL workflows in the following regions:

• US East (N. Virginia, Ohio)
• US West (Oregon)
• Europe (Frankfurt, Ireland, London)
• Israel (Tel Aviv)
• Asia Pacific (Seoul, Singapore, Tokyo)

Wrap-up

• AWS HealthOmics WDL workflows can now use resource fallback: an ordered accelerator list plus automatic CPU fallback
• When the preferred accelerator isn't available, HealthOmics automatically switches to the next option without a resubmission
• Per-profile timeouts let you trade off switching speed against the odds of getting your preferred accelerator
• Available across major US, European, Israeli, and Asia Pacific regions

If accelerator availability has ever stalled your production workflows, this update is for you, especially if you're a bioinformatics researcher or infrastructure engineer working with HealthOmics!