What does this calculator do?
IOPS and throughput are two views of the same storage traffic, joined by the average I/O size. A device doing many small operations can be IOPS-bound while moving little data; the same device doing large sequential operations can saturate its bandwidth at a much lower IOPS figure. This calculator multiplies the two to give throughput in both unit ladders — MiB/s using powers of two and MB/s using powers of ten — because storage tools disagree about which they report, and the difference is about 4.9%. Units follow the standards exactly: kB, MB, GB and TB are powers of ten (1 GB = 10⁹ bytes) while KiB, MiB, GiB and TiB are powers of two (1 GiB = 2³⁰ bytes). The two ladders are never mixed, so "1 GB = 1024 MB" is not treated as a universal rule.
Formula
Throughput is simply operations per second multiplied by bytes per operation.
bytesPerSecond = IOPS × averageIOSizeBytes MiB/s = bytesPerSecond ÷ 2^20 MB/s = bytesPerSecond ÷ 10^6
How to use it
Enter the IOPS figure and the average I/O size, choosing the unit carefully: database and filesystem block sizes are normally quoted in KiB (4 KiB, 8 KiB, 64 KiB), while some vendor sheets use kB. The result is what the IOPS and size imply — whether the device, the controller, the HBA or the fabric can actually sustain it is a separate question.
Worked example
10,000 IOPS at an 8 KiB average I/O size: 10,000 × 8,192 = 81,920,000 bytes per second. That is 78.125 MiB/s, or 81.92 MB/s, and about 274.6582 GiB if sustained for a full hour.
Frequently asked questions
Why show both MiB/s and MB/s?
Because tools differ. Many storage utilities report MiB/s (2²⁰ bytes) while vendor datasheets and network tools report MB/s (10⁶ bytes). The same traffic reads as 78.125 in one and 81.92 in the other — showing both prevents a false discrepancy.
Can a device sustain its peak IOPS at any block size?
No. Peak IOPS figures are usually quoted at a small block size such as 4 KiB. At 64 KiB the same device typically delivers far fewer IOPS, because it becomes limited by bandwidth rather than by operation rate.
Is this the throughput my application will see?
Only if the average I/O size is right and the whole path can keep up. Queue depth, controller cache, RAID write penalties, the HBA or fabric, and the filesystem all sit between the device and the application.
Sources
- SNIA Dictionary — vendor-neutral definitions of RAID levels, stripe, mirror, parity, IOPS and raw versus usable capacity.
- NIST — Prefixes for binary multiples (IEC 80000-13): a drive sold as 2 TB (2 × 10¹² bytes) is about 1.82 TiB as reported by most operating systems, which is why raw and usable figures are shown in the unit you choose.
- Nominal equal-drive capacity model used here: RAID 0 = N × S, RAID 1 = S, RAID 5 = (N−1) × S, RAID 6 = (N−2) × S, RAID 10 = (N/2) × S. Filesystem overhead, metadata, hot spares, vendor reservations, compression and deduplication are not modelled unless you enter them.
- Reviewed: 2026-08-10
Related calculators
Last updated: 2026-08-10 · v1.0.0 · Report an error