Backup and Disaster Recovery · #61

Online Replication Backlog Catch-up Calculator

Calculates how long pending replication data takes to drain while new changes continue.

  • Data is processed in your browser and not sent to a server

Tool

Changes not yet transferred to the target.

60–80 % typical on WAN.

New data written during catch-up.

What does this tool do?

After a link outage, initial sync or a large data load, replication falls behind. Production keeps writing meanwhile; if link capacity is only slightly above the change rate, the backlog takes days to drain and RPO breaks. The tool gives the net catch-up speed and time.

How to use it

  1. Enter the accumulated (pending) data.
  2. Enter link speed, efficiency and the average production change rate (GB/hour) during catch-up; press Calculate.

Fields

Pending data
Changes not yet transferred to the target.
Link speed
Bandwidth reserved for replication.
Link efficiency
60–80 % typical on WAN.
Change rate
New data written during catch-up.

Worked example

InputValue
Pending data500
Link speed100
Link efficiency100
Change rate5

Expected result: A 100 Mbit/s link carries 45 GB/h; with 5 GB/h of new changes net 40 GB/h, a 500 GB backlog drains in 12 h 30 min.

During that time the DR copy can be up to 12 hours old. If the RPO is shorter, seed the initial sync on disk or temporarily increase the link.

Second scenario

InputValue
Pending data2000
Link speed50
Link efficiency80
Change rate15

Expected result: 50 Mbit/s at 80 % is 18 GB/h; with 15 GB/h changes net 3 GB/h — a 2,000 GB backlog drains in about 27.8 days.

Method and assumptions

Link capacity (GB/h) = Mbit/s × efficiency × 106 / 8 × 3600 / 109. Net = capacity − change. Time = backlog / net. With net ≤ 0 the backlog never drains.

Using the result and troubleshooting

If the change rate exceeds 70 % of link capacity the system is prone to falling behind permanently; plan compression, WAN acceleration or a bigger link.

Limits and common mistakes

  • Intraday change variation, latency impact on TCP and replication software compression are not modelled.

Frequently asked questions

What is recommended for initial sync?

For large data sets, shipping data on disk (seeding) and then replicating only the differences is fastest.

Related tools and service

Related service: Backup and Disaster Recovery

Systems online tools · See all online tools