Snapshot in. Open archive out.
Run it against an application-consistent storage snapshot or clone — never the live DB. Every table lands in open CSV, Parquet, Avro or Delta: no production load, no second Oracle license, and a copy that stays readable for decades.
//What is verified here
- Reads off-host from the storage replica (mounted snapshot/clone
.dbf) GA - No session, no listener, no load on production — and no license to read it GA
- Export into 15 open formats (incl. Parquet, Delta, Iceberg, Avro, ORC, Arrow) GA
reclaimDB contains no vendor-specific SnapMirror / SRDF / EBS integration. It simply reads the .dbf of the provided replica — i.e. the snapshot is mounted with native storage tooling, then reclaimDB points at it. That is exactly what makes it vendor-neutral. "Scheduled export" = cron / Ansible, not a built-in scheduler.
//Step by step
Mount the snapshot / clone (off-host)
reclaimDB runs against the files of the storage replica — never the live DB. How the snapshot is provisioned (NetApp SnapMirror, Dell SRDF, ZFS, AWS EBS …) is the storage's job; reclaimDB then reads the mounted .dbf.
# Example: a ZFS clone or a mounted snapshot
mount /dev/.../snap_2026-06-19 /mnt/snap
ls /mnt/snap/oradata/*.dbfExtract off-host — no load on production
reclaimdb autoscan /mnt/snap/oradata reclaimdb restore -m meta.json --parquet /archive/2026-06-19/ /mnt/snap/oradata/*.dbf
No session, no listener, no second Oracle license to read.
Write to future-proof open formats
# Data-lake formats for long retention: reclaimdb restore -m meta.json --delta /archive/delta/ /mnt/snap/oradata/*.dbf reclaimdb restore -m meta.json --iceberg /archive/iceberg/ /mnt/snap/oradata/*.dbf
Scheduled via cron — a license- & version-independent archive
Scheduling is an ops pattern (cron / Ansible / CI), not a built-in daemon:
# /etc/cron.d/reclaimdb-archive — daily at 02:00 0 2 * * * oracle reclaimdb --quiet restore -m /etc/meta.json \ --parquet /archive/$(date +\%F)/ /mnt/snap/oradata/*.dbf
//Relevant exit codes
Test this case on your data
Send us the kind of source you actually have — a .dmp, a .dbf, an RMAN backup piece or a storage snapshot. You get a signed evaluation binary, a sample report, and an honest assessment of which paths apply to your case.