How R12.2 applies patches while users stay on the system — and the phase where it most often goes wrong.
ADOP (Application DBA Online Patching) is the R12.2 utility that applies patches to a second, patched file system and database edition while users continue working on the current one. A short cutover then switches users across. It is what allows patching without a long outage, and it is also the main reason R12.2 administration differs from R12.1.
Online patching removed the long maintenance window, which was the main obstacle to keeping an EBS estate current. In exchange it introduced a dual file system and an editioned database, both of which have to be understood and maintained. Estates that never adopted the discipline tend to fall behind on patching precisely because the mechanism feels unfamiliar.
A new patch edition is created from the running edition. The patch file system is synchronised.
Patches are applied to the patch edition. Users are unaffected throughout, which is the point.
The system performs the work that can be done before cutover, shortening the outage.
The only downtime. Services restart and users move to the patched edition.
Obsolete objects from the old edition are removed. Skipping this causes space and performance problems over time.
A team adopts online patching and is pleased with it: patches apply while users work, and cutover takes minutes rather than a weekend. Over several months they apply a dozen patch cycles and skip the cleanup phase each time, because cleanup takes a while and nothing appears to depend on it.
Database growth becomes the first symptom, then performance on editioned objects, then a patch cycle that fails outright because the environment is carrying accumulated editions it was never designed to hold. The fix at that point is a long maintenance window to clean up properly — the exact downtime online patching existed to avoid, now taken all at once and under pressure.
Online patching is a full cycle: prepare, apply, finalise, cutover, cleanup. Treating the first four as the patch and the fifth as optional housekeeping converts a routine into a debt. The discipline is simply to finish the cycle before starting the next one, and to treat a skipped cleanup as an incident rather than a saving.
| Term | What it means |
|---|---|
| Run edition | The file system and schema users are currently on. |
| Patch edition | The parallel copy being patched while users continue working on the run edition. |
| Cutover | The brief downtime where the patch edition becomes the run edition. |
| Cleanup | Removing the old edition after cutover. Skipping it is what causes space problems later. |
| Editioned object | A database object that exists in both editions, which is what makes the parallel approach possible. |
Old editions accumulate. Cleanup is part of the cycle, not an optional tidy-up afterwards.
Custom objects that do not follow the editioning rules break at cutover, which is the worst moment to find out.
Cutover is short but real. It still needs a window and a communication to users.
An abandoned patch cycle must be cleaned up properly, or the next one inherits the mess.
| Metric | What it tells you | What distorts it |
|---|---|---|
| Cutover duration | The actual user-visible downtime, which is the point of the whole mechanism. | |
| Cycles completed vs cleanups run | Any gap here is accumulating debt that will be paid later with interest. | |
| Patch cycle failures | Often traced back to incomplete prior cycles rather than to the patch being applied. | |
| File system and tablespace growth | The dual file system and editions need headroom; running close to full is where cycles fail. |
We build and support the systems this page describes. Start with a free working proof of concept — no cost, no commitment.