Developer tools · Capacity
Backup window and replication lag calculator.
Hours for a backup to finish, and minutes for a lagging replica to catch up.
Inputs
Timing
Enter the values.
How long will a database backup take?
Backup time is data size divided by sustained throughput, plus any overhead you add for compression, verification or retries. Throughput is limited by the slowest of the disk, network and backup tool, so use a rate you have measured end to end rather than the link speed.
A lagging replica catches up only while it applies changes faster than the primary produces them. The backlog is the lag in seconds times the primary's write rate, and the catch-up time is the backlog divided by the difference between apply rate and write rate. If the apply rate is not higher than the write rate the replica never catches up.
Worked example. 500 GB at 200 MB/s is 2,500 seconds, 41.7 minutes; with 10% overhead it is 45.8 minutes. A replica 60 seconds behind a primary writing 20 MB/s has a 1,200 MB backlog; applying at 50 MB/s it gains 30 MB/s on the primary, so it catches up in 40 seconds.
This treats a gigabyte as 1,000 MB and rates as constant. Real backups slow down under load and replica apply rates vary with the workload, so treat the result as a planning figure and test a restore, not only a backup.
Frequently asked questions
How do I estimate backup time for a database?
Divide the data size by the sustained throughput you measure end to end, then add a percentage for compression, verification and retries.
How long will it take a replica to catch up?
Catch-up time is the backlog (lag seconds x primary write rate) divided by apply rate minus write rate. It is only finite if the replica applies faster than the primary writes.
Why does a replica never catch up?
If the replica apply rate is equal to or below the primary write rate, the backlog never shrinks. Reduce write load, speed up apply, or rebuild the replica.
Is a backup that completes the same as a backup that restores?
No. Only a test restore proves the backup works, and restore time is often longer than backup time, so time it separately.
Sources
Checked
No external figures are used: results are arithmetic on the throughput and rates you enter. Use measured values from your own system. Runs entirely in your browser. Nothing you type is sent or stored.
Scope
Planning estimates, not guarantees. Real systems add overhead this model leaves out and behave differently under load. Measure on your own data and load-test before committing to a design or a bill.