Skip to content
Technologystorage

Backup Window Calculator

Whether a backup fits its window, for both the daily incremental and the periodic full. The full is the one that overruns, and it overruns on the night nobody is watching.

Also called: backup time calculator, backup throughput calculator.

TB
%
Mbps
%
Incremental backup time
1.33

1.33 hours for the daily incremental and 44.44 for a full, against a 8 hour window. The full backup overruns by 36.4 hours. It would need 2778 Mbps to fit, against the 500 available. The daily incremental fits comfortably, which is how this gets missed until the first full runs into business hours. At 500 Mbps the system moves 0.23 TB an hour after 50% reduction. Deduplication cuts what is written and sent rather than always what is read, so it helps the target and the network more than the source.

Full backup time
44.44
Data changed a day
0.6
Throughput in TB an hour
0.23
Throughput a full backup would need
2,778
Data written a month
50.57
Whether it fits
The full backup overruns by 36.4 hours. It would need 2778 Mbps to fit, against the 500 available. The daily incremental fits comfortably, which is how this gets missed until the first full runs into business hours.
On throughput
At 500 Mbps the system moves 0.23 TB an hour after 50% reduction. Deduplication cuts what is written and sent rather than always what is read, so it helps the target and the network more than the source.
How this is calculated

A month of backups

DayTypeTB writtenHoursAgainst the window
Day 1Full1044.44Overruns
Day 2Incremental0.31.33Fits
Day 3Incremental0.31.33Fits
Day 4Incremental0.31.33Fits
Day 5Incremental0.31.33Fits
Day 6Incremental0.31.33Fits
Day 7Incremental0.31.33Fits
Method and background

How this is calculated

A daily incremental only moves what changed, so it is comfortable almost everywhere. The full backup is the constraint, and the two are usually sized together as though the daily figure were representative. Deduplication and compression reduce what is written but not always what is read, so a high reduction ratio helps the target and the network more than it helps the source. The page reports the throughput a full backup would need to fit the window, which is the number to take to a capacity conversation, and the monthly volume written, which is the number that decides the storage bill.

the data actually written after deduplication, in bits, over the throughput achieved
D
Data to move
d
Reduction from dedup and compression
R
Throughput

Worked examples

Each of these is asserted on every build. If a change to the engine ever moved one of these answers, the build would fail before the page could print it.

20 TB with a 3% daily change

Data to protect
20 TB
Daily change rate
3%
Backup window available
8
Backup throughput
500 Mbps
Reduction from dedup and compression
50%
Days between full backups
7

Incremental backup time1.33

10 TB after dedup at 0.225 TB an hour

Open this example

enough throughput to fit the window

Data to protect
20 TB
Daily change rate
3%
Backup window available
8
Backup throughput
5,000 Mbps
Reduction from dedup and compression
50%
Days between full backups
7

Incremental backup time0.13

boundary: ten times the throughput brings the full inside the window

Open this example

Method and limits

What it assumes

  • Sustained throughput, where real backups vary with source load and file size.

What it deliberately does not model

  • Millions of small files can make a backup metadata-bound rather than throughput-bound, and none of this arithmetic applies then.
  • The restore time is the number that matters in an incident, and it is not this one.

Formula version 1.0.0 · definition 1.0.0 · United States · Report a problem with this calculator

Frequently asked questions

Why does my full backup overrun when the daily one is fine?
Because they differ by the change rate. At three percent daily change a full backup moves thirty times as much data as an incremental, so a window sized on the daily figure has no chance.
Does deduplication make backups faster?
It reduces what is written and sent, not always what is read from the source. If the source read is the bottleneck, a high dedup ratio changes the storage bill more than the window.