gce/Disk Iops Throughput Utilisation Checks
Product: Compute Engine
Step Type: AUTOMATED STEP
Description
None
Failure Reason
Disk IOPS/Throughput usage is NOT within optimal limits
Failure Remediation
There can be multiple reasons which can cause Disk IOPS/Throughput usage to increase:
-
Application and GuestOS Operations - Unmanaged and untested application workloads can cause the high influx of IOs to the disk and cause IO operations to be queued, causing throttling at disk levels.
-
CPU Starvation - Small instances(with lesser CPUs) may not have enough CPU to serve all I/Os inflight. https://cloud.google.com/compute/docs/troubleshooting/troubleshooting-performance#cpu_and_memory_performance
-
Network Throttling - High sent/received network traffic can cause network throttling, that can also impacts disk operations. https://cloud.google.com/compute/docs/troubleshooting/troubleshooting-performance#network_performance
-
Insufficient Machine Resources - If your machine’s IOPS and throughput limts are not enought to serve your workloads, this can also cause CPU or Disk IOPS/throughput Starvation. https://cloud.google.com/compute/docs/troubleshooting/troubleshooting-performance#storage_performance
To fix this issue:
- Please optmize your application workloads.
- If needed, please add more resources(CPU, Memory) to the VM.
- Please optmize your Disk performance - https://cloud.google.com/compute/docs/disks/optimizing-pd-performance
- If needed, please change your disk type to get better Disk IOPS/throughput limits - https://cloud.google.com/compute/docs/disks/modify-persistent-disk#disk_type