Data validation and repair

Use Repstance to prove that the target holds the same data as the source, before a cutover and while replication runs. This matters most in financial and healthcare systems.

What Repstance does

Compares all tables or a chosen depth, for example 10% as a quick check.
Runs in parallel with the ongoing replication, so the source is not frozen.
The incremental comparison checks only the data changed since the last validation point. The last incremental comparison is part of the cutover.
Differences are repaired with the built-in repair.
A table needs a primary or unique key to be compared.

In numbers

91,000 rows per second compared, 328 million rows per hour (Aurora PostgreSQL target, Repstance release 6, see the Benchmark)
An average 100 GB database is compared in 1.5 to 2.5 hours (an estimate from the measured speed)
A Fintech customer validated 20 TB in 8,000 tables with a 100% deep comparison that ran in parallel with the replication, then compared the changed data daily (about 150 GB a day)
A Healthcare customer uses a 10% comparison depth over 60,000 tables

Both customer cases are on the home page.