How do I tell whether a VPS is hitting disk IOPS or throughput limits?
A VPS disk can have separate limits for operations per second and bytes per second. Small random operations and large sequential transfers stress these limits differently. Guest latency and queueing provide evidence, but the provider must confirm host-side throttling.
Instructions
1. Check the specified storage policy
Identify the attached volume type, performance allocation and any burst rules. Confirm whether multiple disks share a limit and whether the application’s data is on the volume you are inspecting.
2. Observe a bounded workload sample
If sysstat is installed, read existing activity rather than writing a benchmark file to a production disk.
iostat -xz 1 5Review throughput, operations, latency and queueing. The first report may summarise time since boot; use interval reports for the current incident.
3. Correlate application and provider evidence
Record the affected times and device mapping. High latency can result from guest queueing, application synchronisation, shared backend contention or an allocation limit. A high %util value on a virtual or parallel device does not by itself measure the full backend’s capacity.
Check the result
- The observed device corresponds to the application’s actual storage.
- A suspected limit is supported by workload measurements and provider-side policy or telemetry.
Troubleshooting
- The application is slow while disk metrics are quiet
- Investigate network, locks, CPU and application dependencies instead of assuming storage is responsible.
- The provider confirms a performance cap
- Review whether the workload needs a different volume allocation or application I/O pattern before changing the plan.
Official references
Use the documentation for your installed version when a flag, service name or platform feature differs.