Logical-to-network byte SankeyWidth follows relative bytes; labels retain exact values
Unit Converters
Data Transfer Scale Calculator
Build an auditable storage-and-network scaling path from one baseline payload to logical copies, replicas, compressed bytes, deduplicated transmitted bytes, and transfer duration at an explicitly derated link rate.
LOGICAL-TO-PHYSICAL SCALE PATH
Follow one payload through copies, replicas, reduction, and the link
This calculator is for migration, backup, and capacity planners who need to distinguish logical demand from bytes that actually cross a network. Every multiplier has one declared job; the result is a planning quantity, not measured goodput.
Boundary: compression and deduplication are retained fractions, not savings percentages. They apply after instances and replicas. Link efficiency converts nominal bit rate into payload goodput.
Bytes transmitted—
Transfer duration—
Logical fleet—
Replica estate—
Effective payload rate—
Net scale factor—
BYTE JOURNEY
See where the workload grows and where it contracts
Ribbon width follows the current byte quantity at each boundary. It makes a replica multiplier and a reduction ratio visible as different operations instead of burying both inside one factor.
| Boundary | Quantity | Relative to baseline | Interpretation |
|---|
HOW TO USE
Build the scale path in operational order
- Start with one measured or inventoried logical payload and state whether its unit is decimal or binary.
- Count independent instances before applying replicas; a replica is not another source workload.
- Enter retained compression and deduplication fractions from representative jobs, not marketing savings claims.
- Use nominal link rate and a separately evidenced payload-efficiency factor from the intended transfer path.
- Read the stage ledger before the duration: it shows which assumption creates the dominant byte change.
SCALE FUNDAMENTALS
Six boundaries prevent double counting
CALCULATION METHOD
Multiply quantities first; convert bits and time last
L = B × NR = L × CZ = R × rc × rdg = vn × ηt = 8Z / gB is baseline bytes, N instances, C replicas, rc compression retained, rd unique retained, vn nominal bit rate, η payload efficiency, and t seconds.
REPLICATION ACCOUNTING
A replica changes durability cost, not the original source count
Instance expansion answers how many independent workloads exist. Replica expansion answers how many copies cross this boundary. Apply both only when every replica is actually moved. A target system that creates replicas after ingestion should not multiply network traffic by the target replica count.
REDUCTION EVIDENCE
Compression and deduplication depend on scope and ordering
Compression exploits patterns within a stream; deduplication recognizes repeated chunks or objects inside a configured domain. Encrypting first can eliminate deduplication benefit. Record where each ratio was measured, which population it represents, and whether metadata was included.
GOODPUT AND TIME
Nominal link capacity is a ceiling, not an observed payload rate
The efficiency factor derates the link for sustained payload planning. It does not model queue growth, burst limits, disk reads, retries, or competing traffic. When another boundary is slower than the effective link, that bottleneck controls completion time.
DETAILED CALCULATION PROCESS
Default 500 GiB workload reconciliation
| Symbol | Meaning | Default | Unit |
|---|---|---|---|
| B | Baseline payload | 500 GiB | bytes after conversion |
| N | Instances | 12 | count |
| C | Replicas | 3 | count |
| rc | Compression retained | 0.60 | dimensionless |
| rd | Unique retained | 0.80 | dimensionless |
| η | Payload efficiency | 0.85 | dimensionless |
- Convert 500 GiB × 1,073,741,824 B/GiB = 536,870,912,000 B.
- Logical fleet: 500 GiB × 12 = 6,000 GiB.
- Replica estate: 6,000 GiB × 3 = 18,000 GiB.
- After compression: 18,000 × 0.60 = 10,800 GiB.
- After deduplication: 10,800 × 0.80 = 8,640 GiB.
- Effective rate: 10 Gbit/s × 0.85 = 8.5 Gbit/s.
- Transfer time: 8 × 8,640 GiB ÷ 8.5 Gbit/s = 8,731.42 s, or about 2.43 h.
- Reverse check: 8,640 ÷ 500 = 17.28, equal to 12 × 3 × 0.60 × 0.80.
RESULT INTERPRETATION
The transmitted quantity is the decision result; the ledger explains it
A factor above one means fleet and replica growth exceed reduction. A factor below one means reduction outweighs expansion. Treat duration as conditional: if observed completion is longer, compare storage throughput, concurrency limits, quotas, and retry evidence before altering the byte arithmetic.
EVIDENCE TO RETAIN
Freeze scope before approving capacity
Keep the inventory timestamp, logical-size boundary, instance list, replica location, compression sample, deduplication domain and chunk policy, encryption order, nominal link contract, measured goodput window, concurrency, and service-side rate limits.
LIMITS AND EXCLUSIONS
What the scale path does not simulate
- Per-file headers, manifests, block padding, and packet overhead are not itemized.
- Ratios are constant rather than content-dependent distributions.
- Source read, target write, CPU, queue, and API quotas are not modeled.
- Retries and incremental change during migration are excluded.
- The model does not decide where replicas are created.
WORKED DECISION CASES
Two scale paths with different bottlenecks
Backup fan-out
Twelve tenant backups are replicated across regions before compression. The replica multiplier dominates, so validating where replication occurs matters more than changing display units.
Encrypted archive migration
An archive is encrypted before the target deduplication appliance sees it. A historical pre-encryption dedup ratio is invalid; use 1.00 until representative encrypted data provide evidence.
TECHNICAL GLOSSARY
Terms for the migration record
- Logical bytes
- Application-visible data before expansion.
- Replica
- An additional availability or durability copy.
- Compression ratio
- Retained bytes divided by source bytes.
- Deduplication domain
- Population within which repeated chunks are recognized.
- Goodput
- Useful payload delivered per unit time.
- Bottleneck
- The slowest active resource in the path.
- GiB
- 1,073,741,824 bytes.
- Scale factor
- Transmitted bytes divided by baseline bytes.
IMPORTANT NOTE
Do not approve downtime from link rate alone
Use an end-to-end test when the decision affects downtime, recovery objectives, or contractual delivery. This page reconciles declared factors; it cannot discover an omitted bottleneck.
Frequently asked questions
Should replica count include the primary copy?
Use the number of copies crossing the modeled boundary.
Is a 40% compression saving entered as 40?
No. Enter 60%, the retained compressed share.
Can deduplication retained exceed 100%?
No. Put known overhead in the baseline or a separate model.
Why mix decimal Gbit/s with binary GiB?
Network rates use decimal SI prefixes; the page converts explicitly through bits and bytes.
Can this model incremental transfers?
Yes, when the baseline and ratios describe changed data.
Does efficiency include storage speed?
No. Compare source and target storage throughput separately.
Can compression expand data?
This model caps retained fractions at 100%; add known expansion to the baseline.
Why use a Sankey-style ribbon?
It shows where quantity grows or contracts; the ledger provides the exact audit trail.
AUTHORITATIVE BASIS