Backup-window calculator
Can the data you send each day finish overnight? Check daily capacity, spare bandwidth, and how many nights it takes to clear an existing backlog while new data keeps arriving.
Enable JavaScript to calculate. The worked example below is available without it.
Calculation code runs locally in your browser and does not submit these fields. Shared links include your inputs after the # sign; do not share private network details.
Example: 100 GB per night over 100 Mbps
At 80% usable throughput, a four-hour window can transfer 144 GB. Sending 100 GB leaves 44 GB of spare capacity per night. An additional 200 GB backlog needs five windows to clear while the normal 100 GB continues arriving each day. The daily transfer alone takes about 2.78 hours.
Increase daily data to 144 GB and it fits exactly, but an existing backlog never shrinks. At 200 GB a day, the backlog grows by 56 GB each day. Extending the window, increasing usable throughput, or reducing the transfer volume changes that balance.
Use transferred bytes
Enter the bytes the backup needs to send after any compression and deduplication. A database export rewritten every night can have a very different transfer size from a block-level incremental backup. Use transfer logs from representative jobs instead of estimating from the size of the source filesystem.
The existing backlog is additional work, separate from today’s normal transfer. GB and TB are decimal units; GiB and TiB are powers of two. Mbps means millions of bits per second, not megabytes per second.
The calculation
Window capacity in bytes = Mbps × 1,000,000 ÷ 8 × usable percentage ÷ 100 × window hours × 3,600. Spare capacity is that result minus the daily transfer. When spare capacity is positive, backlog recovery takes the backlog divided by spare capacity, rounded up to whole daily windows.
The model puts each day’s data before that day’s single transfer window. It assumes constant throughput, no failed or missed jobs, and no discarded or coalesced backlog. If measured payload throughput already includes overhead, use 100% efficiency to avoid counting the same allowance twice.
Reserve time separately for snapshots, enumeration, verification, retries, and repository maintenance. A transfer finishing on time does not prove the backup is application-consistent or restorable. This tool does not estimate restore time or promise a recovery-point objective.
For the initial seed of an entire dataset, use the first-upload calculator. For network ranges, use the IPv4 subnet calculator.