What is the IOPS limit for rented storage?
In Oracle Cloud Infrastructure (OCI), the IOPS (Input/Output Operations Per Second) limit for rented storage is highly flexible. Unlike other clouds that lock you into a fixed performance tier based on volume size, OCI allows you to change your performance on the fly using a feature called Volume Performance Units (VPUs).
As of 2026, the IOPS limits are broken down into three main categories:
Each individual block volume can be tuned to different performance levels. The more VPUs you assign, the higher the IOPS:
| Performance Tier | VPUs per GB | IOPS per GB | Max IOPS per Volume |
| Lower Cost | 0 | 2 | 3,000 |
| Balanced | 10 | 60 | 25,000 |
| Higher Performance | 20 | 75 | 50,000 |
| Ultra High Performance | 30–120 | Up to 225 | 300,000 |
Pro Tip: The Ultra High Performance (UHP) tier is a standout in the industry. It scales linearly up to 300,000 IOPS for a single volume, making it ideal for the most demanding enterprise databases.
Even if you attach multiple high-speed volumes, you are eventually limited by the "throughput pipe" of the server (compute instance) they are attached to.
Bare Metal Servers: Most modern Bare Metal shapes (like E4, E5, and E6) can handle an aggregate of 1.3 million IOPS across all attached volumes.
Virtual Machines: The limit for VMs is usually capped by the network bandwidth. Most flexible VMs can reach up to 600,000 aggregate IOPS, provided you have enough OCPUs to handle the traffic.
If you rent a Dense I/O bare metal shape, you aren't using network-attached block storage. You are using physical NVMe drives bolted directly to the server.
Performance: These drives can often exceed several million IOPS with microsecond latency, but remember: this storage is ephemeral. If you delete the server, the data goes with it.
Dynamic Scaling: You can drag a performance slider in the OCI console to increase IOPS from 3,000 to 300,000 without rebooting your server or detaching the volume.
Auto-Tuning: You can enable "Performance Auto-tune" so OCI automatically cranks up the IOPS when it detects a heavy load (like a month-end report) and drops it back down to save you money when the server is idle.
SLA Guarantee: Oracle is one of the only major cloud providers that offers a Performance SLA, meaning they financially back their promise to deliver these IOPS consistently.