IOPS / MBps / GB-per-day Converter
Convert IOPS, throughput, and daily volume from I/O size and active hours, then check reserve-adjusted ceilings and cloud disk limits.| Measure | Value | Meaning | Copy |
|---|---|---|---|
| {{ row.label }} | {{ row.value }} | {{ row.note }} |
| Reference profile | Required setting | Published maximum | Fit | Copy |
|---|---|---|---|---|
| {{ row.label }} | {{ row.required }} | {{ row.maximum }} | {{ row.fit ? 'Within limits' : 'Exceeds limit' }} |
| Decision | Current reading | Next check | Copy |
|---|---|---|---|
| {{ row.label }} | {{ row.reading }} | {{ row.action }} |
A backup that must move 1 TB overnight does not have the same rate requirement as 1 TB spread evenly across a day. If the useful window is four hours, the storage and network path must sustain about 69.44 MB/s during those four hours. The calendar-day average hides that demand.
Storage planning commonly moves between three quantities. Input/output operations per second (IOPS) counts requests. Throughput counts bytes moved each second. Daily volume counts all bytes moved during a chosen window. Average I/O size connects them because one operation may carry 4 KiB, 64 KiB, 1 MiB, or another payload.
- IOPS
- Operation rate. It says how many requests finish each second, not how much data each request carries.
- Throughput
- Payload byte rate. MB/s and MiB/s are byte units; Mbps and Gbps are bit units.
- Daily volume
- Total bytes delivered during the selected active hours, which may be much shorter than 24 hours.
- Headroom
- Capacity left below a reserve-adjusted ceiling after the calculated payload rate is applied.
Workload shape determines which limit becomes tight first. Small random requests can exhaust an IOPS allowance while leaving throughput available. Large sequential requests can fill a throughput ceiling with far fewer operations. The average block size should therefore come from storage telemetry when possible, not from a generic workload label.
| Unit | Bytes represented | Common reading risk |
|---|---|---|
| 1 MB/s | 1,000,000 B/s | Decimal byte rate |
| 1 MiB/s | 1,048,576 B/s | Binary byte rate |
| 1 Mbps | 125,000 B/s | Bits are divided by 8 to obtain bytes |
| 1 GiB | 1,073,741,824 B | Not equal to 1 GB |
A converted payload rate is a capacity-planning estimate. It does not model latency, queue depth, read/write mix, caching, compression, protocol overhead, host limits, burst credits, or contention. Those checks remain necessary before choosing a disk, instance, or link.
How to Use This Tool:
Begin with the quantity you actually measured or must deliver, then keep its units and time window explicit.
- Choose IOPS, sustained throughput, or daily volume under Starting metric, then enter its value and matching unit.
- Enter the observed Average I/O size or choose the closest workload preset. This value directly links operation rate and byte rate.
- Set the Active delivery window to 24 hours for continuous load or to the actual backup, ingest, or migration window.
- Optionally choose a nominal link or custom sustained Ceiling comparison, then apply a measured or policy-based Safety reserve.
- Review the converted IOPS, active-window throughput, moved volume, block-size pressure, and any reserve-adjusted utilization before comparing cloud profiles.
Interpreting Results:
The active-window throughput is the rate that must be sustained while the job runs. The 24-hour average is useful for reporting, but it can understate the capacity needed for a short delivery window.
- Confirm average I/O size from telemetry because doubling bytes per operation doubles throughput at fixed IOPS and halves IOPS at fixed throughput.
- A reserve-adjusted ceiling load above 80% is a warning; above 100% means the calculated payload rate exceeds the safe ceiling.
- Negative headroom means the safe ceiling is already exceeded.
- A cloud profile marked within limits only passes the stored headline IOPS and throughput rules. Recheck volume size, region, attachment, instance, and service-specific limits in current provider documentation.
Technical Details:
Every source quantity is normalized to bytes, bytes per second, or operations per second before conversion. Decimal prefixes use powers of 1000, binary prefixes use powers of 1024, and bit-rate units are divided by 8. Full floating-point precision is kept through the calculation; rounding belongs to display formatting.
Formula Core
Average bytes per operation is the bridge between IOPS and active throughput.
Moved volume uses the active hours rather than assuming a continuous day. The reverse calculation gives the sustained rate required to deliver a known volume.
A selected ceiling is reduced by the safety reserve before utilization and headroom are calculated.
| Symbol | Meaning | Domain |
|---|---|---|
| R | Active payload rate in B/s | Selected source value must produce a rate greater than zero. |
| B | Average bytes per operation | Greater than zero and no more than 1 GiB. |
| h | Active delivery hours per day | 0.25 to 24 hours, inclusive. |
| C | Selected nominal or custom ceiling in B/s | Zero disables comparison. |
| p | Safety reserve percentage | 0% to 50%, inclusive. |
Rule and Lookup Core
Ceiling badges use strict upper comparisons. Exactly 80% is within the normal range, more than 80% through 100% is a warning, and more than 100% is over the reserve-adjusted ceiling.
| Safe utilization | Status | Meaning |
|---|---|---|
| 80% or less | Within reserve | At least 20% of the safe ceiling remains. |
| Greater than 80% and at most 100% | Review | Reserve-adjusted headroom is small but non-negative. |
| Greater than 100% | Exceeds safe ceiling | Calculated headroom is negative. |
Cloud fit applies the current headline caps and the provider coupling rules represented below. These are single-volume reference checks, not quotes or provisioning guarantees.
| Reference profile | Derived requirement | Stored headline fit boundary |
|---|---|---|
| AWS EBS gp3 | The greatest of measured IOPS, 3,000, or the greater of workload MiB/s and 125 divided by 0.25 | Required IOPS at most 80,000 and workload throughput at most 2,000 MiB/s |
| AWS EBS io2 Block Express | The greater of measured IOPS or workload MiB/s divided by 0.256 | Required IOPS at most 256,000 and throughput at most 4,000 MiB/s |
| Azure Ultra Disk | The greatest of measured IOPS, 100, or workload MB/s divided by 0.25 | Required IOPS at most 400,000 and throughput at most 10,000 MB/s |
| Google Hyperdisk Throughput | The greatest of workload MiB/s, derived IOPS divided by 4, or 20 MiB/s | Derived IOPS at most 9,600 and required throughput at most 2,400 MiB/s |
Limitations:
The result is a deterministic payload conversion based on one average I/O size and one active window. It is not a latency, queueing, or complete infrastructure model.
- Protocol framing, filesystem and database overhead, retries, compression, deduplication, caching, and read/write mix are not subtracted or modeled.
- Nominal 1, 10, and 25 Gbps link presets do not guarantee equivalent sustained payload throughput.
- Cloud limits can depend on volume size, machine type, attachment mode, region, quotas, and newer provider changes.
- All calculations stay in the browser and no workload values are sent for conversion.
Worked Examples:
Small-block database load
At 10,000 IOPS with an 8 KiB average operation, active throughput is 81.92 MB/s or 78.125 MiB/s. Running that rate for an 8-hour window moves 2,359,296,000,000 bytes; the same IOPS with 64 KiB operations would create eight times the throughput.
Four-hour data move
Moving 1 TB in four hours requires about 69.44 MB/s. With a 64 KiB average I/O size, that rate is about 1,059.64 IOPS. A 24-hour average would be much lower and would understate the rate needed during the job.
References:
- Definitions of binary prefixes, National Institute of Standards and Technology.
- Amazon EBS General Purpose SSD volumes, AWS Documentation.
- Amazon EBS Provisioned IOPS SSD volumes, AWS Documentation.
- Select a disk type for Azure IaaS VMs, Microsoft Learn.
- About Hyperdisk Throughput, Google Cloud Documentation.