Data Transfer Rate Converter
Convert data transfer rates between bits, bytes, SI, and IEC units, model overhead, and test payload ETAs against a time window.{{ result.summaryTitle }}
Conversion result
| Equivalent rate | Line side | Payload side | Copy |
|---|---|---|---|
| {{ row.label }} | {{ row.lineDisplay }} | {{ row.payloadDisplay }} |
| Payload checkpoint | Payload ETA | Line ETA | Window fit | Copy |
|---|---|---|---|---|
| {{ row.label }} | {{ row.payloadEta }} | {{ row.lineEta }} | {{ row.windowStatus }} |
| Metric | Value | Copy |
|---|---|---|
| {{ row.label }} | {{ row.value }} |
| Metric | Value | Copy |
|---|---|---|
| {{ row.label }} | {{ row.value }} |
A transfer estimate can be wrong before any packet or file moves if the number loses its unit. 1 Gbps and 125 MB/s describe the same decimal line speed before overhead, but they are written for different audiences. Network links are usually quoted in bits per second because physical and logical links transmit bits. File copies, backup tools, and storage counters often report bytes per second because payloads are stored and counted in bytes.
The bit-to-byte bridge is fixed at eight bits per byte. Prefixes are less forgiving. Decimal SI prefixes scale by powers of 1000, so GB and Gbps use the same decimal ladder. Binary IEC prefixes scale by powers of 1024, so GiB and Gibps belong to a different ladder. At small sizes the difference is easy to miss; at backup, migration, and media-ingest scale it can change an ETA by minutes or hours.
Overhead is the other part of the estimate. Ethernet framing, TCP/IP headers, encryption, VPN encapsulation, retries, and application behavior can reserve part of a line for data that is not the payload being copied. A line rate is the raw path speed before that reserve. Payload throughput is the useful data rate after it. Treating one as the other makes a tight window look safer than it is.
| Term | Plain meaning | Planning effect |
|---|---|---|
| Line rate | Raw link, port, bus, or path speed before overhead is removed. | Useful for link sizing, vendor specs, and standard Ethernet checkpoints. |
| Payload throughput | Useful bytes moved after headers, framing, encryption, or tunnel reserve. | Closer to backup, copy, ingest, and application transfer time. |
| Observation window | The allowed time for a reference payload to finish. | Turns a rate conversion into a pass or miss schedule check. |
| Notation shift | The change from reusing the same number with an SI or IEC paired label. | Prevents GB, GiB, Gbps, and Gibps from being treated as interchangeable. |
Arithmetic can give a clean planning baseline, but it cannot certify a live path. Latency, congestion, packet loss, retransmits, disk speed, CPU encryption cost, compression, deduplication, single-stream limits, and throttling can all change observed throughput. When a backup window or migration cutover has real cost, leave margin and test with representative data.
How to Use This Tool:
Enter the rate as it was reported, then decide whether that number is raw capacity or already measured useful throughput.
- Enter Input rate and choose the matching unit, such as
Mbpsfor a link speed orMB/sfor a file-copy readout. The summary headline updates to the selected target unit. - Set Rate basis to Line rate for a port speed, circuit, bus, or vendor specification. Choose Payload throughput for an application counter, copy dialog, or monitoring graph after overhead has already affected the number.
The basis changes the overhead direction: line inputs lose overhead before payload ETA, while payload inputs estimate the line rate needed to produce that useful speed.
- Choose Convert to for the unit you need to quote first. File planning usually reads better in
MB/sorMiB/s, while link sizing usually reads better inMbpsorGbps. - Select an Overhead profile or choose Custom and enter Protocol overhead from 0% to 40%.
A 0% overhead model is useful for raw storage or already measured application throughput, but it is usually too optimistic for tunneled or encrypted network transfers.
- Open Advanced when the job has a real payload and deadline. Set Capacity labels to SI for decimal labels or IEC for KiB, MiB, GiB, and TiB paths.
- Enter Reference payload, Observation window, Active hours per day, and Billing month length when you need ETA, window fit, daily payload, or monthly payload estimates.
- Check Window Target for the pass or miss verdict, then use Rate Ledger, Checkpoint ETAs, and Conversion Brief to copy the exact row or export a handoff table with the unit, basis, and overhead context intact.
Interpreting Results:
The headline is the converted value in the selected target unit for the active basis. The supporting text keeps the other basis visible, so a line-rate input also shows modeled payload throughput, and a payload-throughput input also shows the estimated line equivalent.
Rate Ledger is the best check for bit-versus-byte equivalents. Checkpoint ETAs compares common payload sizes against the chosen window. Window Target is the schedule check for the reference payload. Conversion Brief gathers basis, overhead, standard fit, notation warning, payload budgets, and action text for review.
| Result cue | How to read it |
|---|---|
| Equivalent line rate | Raw capacity implied by the current settings before overhead is removed. |
| Equivalent payload rate | Useful data rate after the overhead reserve is applied. |
| Window verdict | Whether the reference payload clears the selected time budget, with slack or miss time. |
| Required line rate | Raw path speed needed to hit the window after overhead is included. |
| Same-number notation shift | How much the value changes if the same number is reused with the paired SI or IEC label. |
| Current standard fit | The smallest built-in Ethernet checkpoint that covers the modeled line rate. |
A pass verdict means the arithmetic clears the selected window under the current assumptions. It does not mean the real network or storage path will sustain that speed. If the pass is narrow, compare Current payload rate, Required payload rate, and Comfort tier, then add margin or test the path under load.
Technical Details:
Transfer-rate conversion starts by reducing the entered value to bits per second. Bit units stay in bits. Byte units multiply by eight. Decimal units use powers of 1000, and binary units use powers of 1024. After that normalization, each displayed rate is the same underlying quantity divided by the target unit factor.
Overhead changes whether the entered value is reduced or expanded. A line-rate input models useful payload throughput by subtracting overhead. A payload-throughput input already represents useful data, so the corresponding line rate is estimated by adding the reserve back. The same overhead percentage therefore has opposite direction depending on the selected basis.
Formula Core
The equations use the unrounded model values. Display rows round values for readability, but ETA, required-rate, and window-fit decisions are calculated from the normalized numbers.
| Symbol or term | Meaning | Related input or result |
|---|---|---|
| v | Entered numeric rate. | Input rate |
| Funit | Bits-per-second factor for the selected unit label. | Rate Ledger |
| Rline | Raw path speed before overhead. | Equivalent line rate, Current line rate |
| Rpayload | Useful payload speed after overhead. | Equivalent payload rate, Current payload rate |
| h | Protocol overhead percentage, from 0% to 40%. | Protocol overhead, Overhead reserve |
| twindow | Observation window converted to seconds. | Allowed window, Window verdict |
For a line-rate example, 1 Gbps normalizes to 1,000,000,000 bits per second. With 3% overhead, payload throughput is 970,000,000 bits per second. In decimal byte units that is 121.25 MB/s, so a 100 GB reference payload takes about 825 seconds, or 13m 45s.
Unit Families
| Family | Examples | Scaling rule | Common use |
|---|---|---|---|
| SI bits | Kbps, Mbps, Gbps, Tbps | 1000 per prefix step | Network links, circuits, and Ethernet speed classes. |
| IEC bits | Kibps, Mibps, Gibps, Tibps | 1024 per prefix step | Binary-labeled rate comparisons. |
| SI bytes | KB/s, MB/s, GB/s, TB/s | 8 bits per byte, then 1000 per step | File-copy, vendor storage, and decimal payload readouts. |
| IEC bytes | KiB/s, MiB/s, GiB/s, TiB/s | 8 bits per byte, then 1024 per step | Binary storage paths, memory-oriented counters, and some bus figures. |
Payload sizes use the selected size label. MB, GB, and TB are powers of 1000. MiB, GiB, and TiB are powers of 1024. Daily and monthly payload budgets multiply payload bits per second by active seconds, then divide by eight to return bytes before formatting the size label.
Planning Boundaries
| Area | Rule or range | Interpretation limit |
|---|---|---|
| Protocol overhead | 0% to 40%, with presets for common planning cases. | Packet loss, retransmits, congestion, and disk waits are not modeled separately. |
| Reference payload | Positive MB, GB, TB, MiB, GiB, or TiB value. | Compression, sparse files, dedupe, and filtering can change real bytes moved. |
| Observation window | Seconds, minutes, or hours converted to total seconds. | A barely passing window needs operational margin. |
| Active delivery budget | Active hours per day from 0.1 to 24, and month length from 1 to 31 days. | Calendar output is arithmetic capacity, not a provider billing rule. |
| Ethernet checkpoint fit | Built-in tiers from 100 Mb Ethernet through 400 Gb Ethernet. | A tier fit is a sizing clue, not a cable, optic, host, switch, or protocol guarantee. |
The same payload-throughput model drives the checkpoint ETA rows, the current window capacity, the required payload rate, the required line rate, and the active-day or month payload budgets. Comparisons are fairest when the rate basis, overhead percentage, SI or IEC label choice, and active-hour window stay fixed across runs.
Accuracy and Privacy Notes:
The calculations are deterministic unit and rate arithmetic based on the values entered on the page. They do not measure the current browser connection, storage path, remote service, or live network speed.
Rate, payload, overhead, and window values do not need to be uploaded for the conversion itself. Copy, CSV, DOCX, and JSON actions create handoff artifacts from the current result. Avoid adding confidential project names, customer identifiers, or internal file paths to exported context if the output will be shared outside the team.
Use measured throughput from the real path whenever a schedule has cost. A modeled 20% headroom tier is usually safer than a narrow pass, especially across VPNs, long-distance links, encrypted transfers, and shared storage.
Worked Examples:
1 Gb Ethernet copy estimate
Enter 1 Gbps as Line rate, convert to MB/s, and choose the Ethernet TCP payload profile at 3%. The line rate is 125 MB/s, payload throughput is 121.25 MB/s, and a 100 GB reference payload finishes in about 13m 45s. A 1 hour window passes with wide margin.
VPN graph reported as payload
A monitoring graph reports 250 Mbps after the tunnel overhead has already affected traffic. Choose Payload throughput, use the VPN or tunneled WAN profile at 12%, and convert to MB/s. The payload rate is 31.25 MB/s, while the estimated line equivalent is about 35.51 MB/s.
Short maintenance window miss
Enter 100 Mbps as a line rate with 3% overhead, then test a 10 GB payload inside 10 minutes. Payload throughput is 12.125 MB/s, so the ETA is about 13m 45s. Window Target misses by about 3m 45s and shows the required payload and line rates for the same payload.
Binary storage label check
A storage path quoted at 3.5 GiB/s should keep IEC labels when comparing with a TiB-sized payload. With 0% overhead, it converts to 28 Gibps. A 1 TiB payload inside 5 minutes may clear arithmetically, but the narrow margin still needs a real copy test.
FAQ:
Why does Mbps not match MB/s?
Mbps is megabits per second. MB/s is megabytes per second. One byte is eight bits, so 100 Mbps equals 12.5 MB/s before overhead.
Should I use line rate or payload throughput?
Use Line rate for raw link, port, bus, or vendor speeds. Use Payload throughput when the number came from a file-copy dialog, application metric, or monitoring graph that already reports useful traffic.
Why do SI and IEC labels change the result?
SI prefixes use powers of 1000, while IEC prefixes use powers of 1024. The Same-number notation shift row shows how much changes when the same numeric value is paired with the other label family.
Why does the window pass in the calculator but fail in real life?
The window verdict models unit conversion and overhead only. Congestion, latency, retransmits, disk limits, encryption CPU, throttling, compression, and application behavior can reduce observed throughput.
Can I use the result as a transfer cap?
Yes, if the receiving command accepts a rate in the same unit. Convert the desired ceiling to the command's expected label, then test the achieved average because client-side rate limits often settle over several seconds.
Glossary:
- Bit
- The smallest binary data unit, commonly used for network and link rates.
- Byte
- A group of eight bits, commonly used for files, buffers, and storage payloads.
- SI prefix
- A decimal prefix such as kilo, mega, giga, or tera that scales by powers of 1000.
- IEC prefix
- A binary prefix such as kibi, mebi, gibi, or tebi that scales by powers of 1024.
- Line rate
- Raw path speed before protocol or transfer overhead is removed.
- Payload throughput
- Useful data movement after overhead has been accounted for.
- Observation window
- The time budget used to decide whether a reference payload can finish.
References:
- Metric (SI) Prefixes, National Institute of Standards and Technology.
- Prefixes for Binary Multiples, National Institute of Standards and Technology.
- IEEE Standard for Ethernet, 100 Gb/s, 200 Gb/s, and 400 Gb/s Operation, IEEE Standards Association, 2022.
- How to limit bandwidth in cURL, Simplified Guide.