Calculate disk IOPS from seek time, rotational latency, and transfer time. Compare HDD and SSD random I/O performance.
IOPS (Input/Output Operations Per Second) is the fundamental measure of storage device performance for random access workloads. For HDDs, IOPS is primarily limited by mechanical factors: the time the read/write head takes to move (seek time) and the time for the platter to rotate to the right position (rotational latency). A typical 7,200 RPM enterprise HDD delivers about 75–100 random IOPS, while a 15,000 RPM drive might reach 180 IOPS.
SSDs eliminate mechanical latency entirely, delivering tens of thousands to hundreds of thousands of random IOPS. This dramatic difference is why SSD adoption has been transformative for database and virtualization workloads. This calculator lets you model IOPS from the physical characteristics of any storage device, showing the theoretical maximum for both random reads and writes.
Integrating this calculation into monitoring and reporting workflows ensures that engineering decisions are grounded in real data rather than assumptions about system behavior. Precise measurement of this value supports informed infrastructure decisions and helps engineering teams optimize system architecture for both performance and cost efficiency.
IOPS is the key performance metric for databases, virtual machines, and any workload with random access patterns. This calculator derives IOPS from physical drive characteristics, helping you compare devices, validate vendor claims, and size storage for performance requirements. Precise quantification supports capacity planning and performance budgeting, ensuring infrastructure investments are right-sized for both current workloads and projected future growth.
IOPS = 1000 / (seek_time_ms + rotational_latency_ms + transfer_time_ms)
Result: 118 IOPS
For a 7,200 RPM HDD: seek time 4.2 ms + rotational latency 4.17 ms (half rotation at 7200 RPM) + transfer time 0.1 ms = 8.47 ms per IO. 1000 / 8.47 = 118 IOPS. An SSD with 0.05 ms seek, 0 ms latency, and 0.01 ms transfer yields 16,667 IOPS—over 140× faster.
HDD IOPS is dominated by seek time and rotational latency. Seek time varies from 0.5 ms (short stroke) to 12 ms (full stroke), averaging 3–5 ms for enterprise drives. Rotational latency of 4.17 ms (7200 RPM) or 2.0 ms (15000 RPM) is the other major component. Transfer time is negligible for small IOs.
SSDs deliver IOPS through parallelism across multiple NAND channels and dies. NVMe SSDs connect via PCIe, eliminating the AHCI overhead of SATA. A modern NVMe drive can sustain 500,000+ random read IOPS at queue depth 256. Write IOPS are typically 60–80% of read IOPS due to garbage collection overhead.
To size storage for a database, multiply the expected transactions per second by the average IOs per transaction (typically 5–20). Add 20% overhead for background tasks. This gives the minimum IOPS your storage must deliver.
It depends on the workload. A single desktop user needs 50–100 IOPS. A busy database server may need 5,000–50,000 IOPS. Virtualization hosts typically need 100–500 IOPS per VM. Enterprise NVMe SSDs can deliver 100,000–1,000,000+ IOPS.
Deeper queues allow the device to optimize IO scheduling and serve multiple requests in parallel. An SSD at queue depth 1 might deliver 10,000 IOPS, but at queue depth 32 it might reach 100,000+ IOPS. HDDs benefit less from queue depth.
SSDs must erase blocks before writing (erase-before-write), which adds latency. HDDs need write verification. For SSDs, the controller's garbage collection and wear leveling also consume write bandwidth, reducing sustained write IOPS.
Random IOPS measures performance when accessing scattered locations on the drive. Sequential IOPS (or throughput) measures performance when reading/writing contiguous data. HDDs have huge differences between the two; SSDs perform similarly for both.
Throughput (MB/s) = IOPS × block_size (KB) / 1024. For 4K random IO at 100,000 IOPS: 100,000 × 4 / 1024 = 390 MB/s. For 128K sequential IO at 5,000 IOPS: 5,000 × 128 / 1024 = 625 MB/s.
Yes. RAID 0 and RAID 10 can multiply read IOPS by the number of drives. RAID 5 improves read IOPS but write IOPS suffer due to parity calculations. RAID 6 has even more write penalty. RAID 1 doubles read IOPS for two drives.