1. Intent
Prefect
Deployment: rnaseq-analysis/prod
Flow: rnaseq_analysis
Flow run: #4521
Task: STAR_ALIGN
Work pool: aws-batch-pool

Requested CPUs: 64
Requested memory: 488 GB
2. Execution (Prefect Worker / Infrastructure)
Execution Reality
Prefect worker: aws-batch-worker
Infrastructure block: rna-highmem
Job: AWS Batch job (ID 4521)
Instance: r5.16xlarge
Container image: ../biocontainers/star
Task run: STAR_ALIGN
Duration: 47 min
Prefect submitted the task using the deployment's infrastructure configuration. Runtime behavior was not observable from Prefect logs or infrastructure metrics alone.
3. Reality
Observed inside Prefect task run STAR_ALIGN
Prefect deployment rnaseq-analysis → flow run #4521 → task run STAR_ALIGN
CPU
8%
Memory
18%
Disk I/O
114%
Diagnosis: Task run is I/O-bound. Increasing CPU or memory in the Prefect infrastructure block will not improve runtime.
4. The Fix
Recommended Prefect infrastructure override for task STAR_ALIGN:
Infrastructure profile: rna-io-optimized
Requested CPUs: 8
Requested memory: 64 GB
Storage: Local NVMe scratch
Instance type: r6id.2xlarge
Scope: Task-level override (not global deployment)
This change updates the Prefect deployment's run configuration. Same flow, same task — ~85% lower compute cost and 3.7× faster runtime.