Free features sit one keyword away from licensed ones
Advanced Compression is the database option most likely to be running without a single deliberate decision behind it. It is not an installable product, it is a property of objects and settings, so nothing stops an unlicensed database from using it. Whoever writes the schema, clones an environment, or copies a backup configuration can commit the estate, and the price is set at the engine scale.
Prepared by Redress Compliance · August 11, 2026 · Oracle advisory. Based on 20 to 26 Oracle compression reviews, 2024 and 2025.
Executive summary
In 3 out of 4 estates the licensed compression had been set by accident, usually a migration tool default. That single figure explains why this option produces findings out of proportion to its usefulness: nobody chose it.
A Data Pump import recreates the compression attributes of the source schema, so one licensed source seeds a dozen unlicensed targets, and a tablespace created with a compressed default silently compresses every table later created inside it.
The option gates ten feature families, not just table compression, and most buyers know only the first.
Row, index, and large object compression sit alongside backup compression levels, export compression, standby redo transport compression, network compression, heat map with automatic data optimisation, and archive optimisation.
Each is invoked by syntax that looks like tuning rather than purchasing, which is precisely why the exposure accumulates without a decision trail behind it.
The free features sit one keyword away from the licensed ones in the same statement.
Basic table compression is free while the advanced variant bills; the default backup algorithm is free while the higher levels bill; metadata only export compression is free while full or data only bills; the storage format itself is free while compression and deduplication on top of it are not.
That proximity is the mechanism, because a hardening guide or a copied script changes one word and commits the estate.
The option must match the database licence in metric and quantity, so one compressed object prices every licensed processor under that database. At roughly 11,500 dollars per processor with 2,530 dollars annual support, the same price line as Partitioning, a finding is never small.
Reverting accidental compression and licensing only where it genuinely earns its keep cut the proposed finding by 18 to 30 percent in our reviews, and fewer than 1 in 3 accidentally compressed objects delivered measurable benefit at all.
Where the free line sits, feature by feature
| Area | Free | Licensed |
|---|---|---|
| Table compression | Basic, for direct path loads | The advanced row variant |
| Index compression | The older prefix style | The advanced variants |
| Backup compression | The default basic algorithm | The higher compression levels |
| Export compression | Metadata only | Full or data only |
| Large object storage | The storage format itself | Compression and deduplication on top of it |
| Columnar compression | Included on engineered storage | Not part of this option at all |
Two structural facts explain why this option produces findings more reliably than almost any other.
It is invoked by syntax that looks harmless, so the decision never reaches anyone with a licensing responsibility, and it is priced at the same line as Partitioning, so the finding is never small enough to absorb quietly.
The edition boundary adds a third: the option exists only on Enterprise Edition, which means a compressed attribute travelling into a Standard Edition database during a downgrade project is a configuration error to fix rather than a licence to purchase.
Cloud shapes change the answer again, since some service tiers bundle the option while a bring your own licence move carries the on premises entitlement question with it, so confirm which you are on before paying twice. The wider option exposure sits in the options pricing and audit guide.
The five accidental paths in
- Imports carry attributes. An export and import recreates the compression attributes of the source schema, so one licensed source seeds a dozen unlicensed targets without anyone choosing to.
- Tablespace defaults cascade. A tablespace created with a compressed default silently compresses every table later created in it, which makes one line of setup a standing exposure.
- Vendor scripts decide for you. Some third party installers and schema upgrade kits ship compressed definitions, and nobody on the buyer side reads the DDL.
- Copied backup configuration. A script cloned from another environment brings the higher compression algorithm with it, and the setting persists without announcing itself.
- Standby parameter drift. A latency fix that enables redo transport compression turns a network tweak into a licensing decision, and a standby must mirror the primary's option licensing anyway.
The Oracle CIO complete playbook
The option exposure, the evidence discipline, the audit sequence, and the five year plan to control Oracle spend.
Get the white paper →Detecting it, and the remediation that holds
Detection is straightforward once you know the ten places to look, and the usual failure is looking in one of them.
Table and index attributes appear in the data dictionary views for tables, indexes, and large objects; backup and export usage appears in the feature usage statistics and in the backup configuration itself.
Redo transport and network compression appear in instance parameters and network configuration files rather than anywhere a database query would find them; and the heat map and archive optimisation features appear in their own parameter and policy views.
An audit that queries only for compressed tables will miss six of the ten families, which is how estates arrive at a finding convinced they had checked.
One measurement caveat is worth knowing before conceding anything, because it cuts in the buyer's favour: running the compression advisor creates temporary compressed objects while it estimates ratios, and those runs can leave entries in the usage statistics.
The vendor's own scripts attempt to separate advisor activity from production use, so establish which one a flagged entry reflects rather than accepting the raw statistic. Remediation is legitimate and it is the whole defence.
Revert the attributes where the compression was accidental, document the dates on which each change was made, keep the before and after evidence together.
And license only where compression genuinely earns its keep, which in our reviews was fewer than one in three of the accidentally compressed objects.
Migration projects are the usual patient zero, because attributes move unseen, the estate goes live, and the project team disbands before anybody sweeps for feature usage.
A post migration licence verification run in the week after go live is the cheapest control available and almost nobody schedules it. The wider exposure register sits in hidden Oracle audit exposures.
- Percentile standing for your exact deal size and industry, from real closed transactions
- Scenario simulation before the call: test alternative terms and see the financial impact of each
- A negotiation playbook, talking points, and a two page executive brief on day one
What we saw across compression reviews, 2024 and 2025
In the 20 to 26 Oracle compression reviews run through 2024 and 2025, accidental usage was the dominant pattern rather than an occasional one:
Estates where licensed compression had been set without a deliberate decision, usually through a migration tool default.
Cut achieved by reverting accidental compression and licensing only where the compression genuinely earned its keep.
Three patterns recurred: licensed compression set by accident in 3 out of 4 estates, usually a migration tool default; the accidentally compressed objects delivering measurable benefit in fewer than 1 in 3 cases.
And reverting the accidental use while licensing only where needed cutting the proposed finding by 18 to 30 percent.
The buyer side move is to sweep all ten feature families rather than only the tables, separate advisor artefacts from production usage before conceding anything, revert what was never chosen, and schedule a post migration verification in the week after any go live.
The wider library sits in the Oracle practice.
Your first five moves
- Sweep all ten feature families, not just compressed tables, because six of them live in instance parameters, backup configuration, and network files rather than in the data dictionary.
- Separate advisor artefacts from production usage before conceding anything, since running the compression advisor creates temporary objects that can leave entries in the usage statistics.
- Revert what was never chosen and date every change, keeping the before and after evidence together, because remediation is legitimate and the dates are what make it defensible.
- Check tablespace defaults and vendor supplied definitions, which are the two paths that convert a single setup line into a standing estate wide exposure.
- Schedule a licence verification for the week after any migration go live, the cheapest control in the category and the one almost nobody runs. The Oracle practice runs the sweep with you.
Frequently asked questions
What does Advanced Compression actually cost?
Roughly 11,500 dollars per processor with about 2,530 dollars annual support, the same price line as Partitioning.
Because the option must match the database licence in metric and quantity, a single compressed object prices every licensed processor under that database, which is why a finding here is never small.
What does the option gate beyond table compression?
Ten feature families in total: advanced row, index, and large object compression, large object deduplication, the higher backup compression levels, full export compression, standby redo transport compression, network compression, heat map with automatic data optimisation, and archive optimisation.
Most buyers know the first and are caught by the other nine.
Where does the free line sit?
One keyword away in the same statement.
Basic table compression is free while the advanced variant bills, the default backup algorithm is free while the higher levels bill, metadata only export compression is free while full or data only bills.
And the large object storage format is free while compression and deduplication on top of it are not.
How does accidental use happen?
Through five paths: imports recreating source schema attributes, tablespace defaults cascading to every table created in them, third party installers shipping compressed definitions, backup scripts cloned with a higher algorithm, and standby parameter changes made for latency.
In 3 of 4 estates the compression had been set with no deliberate decision behind it.
Is remediation legitimate?
Yes, and it is the whole defence. Revert the attributes where compression was accidental, document the dates of each change, keep the before and after evidence together, and license only where compression genuinely earns its keep.
Doing so cut the proposed finding by 18 to 30 percent across our reviews.
Does the compression advisor create a false positive?
It can. Running the advisor creates temporary compressed objects while estimating ratios, and those runs can leave entries in the usage statistics.
The vendor's own measurement scripts attempt to separate advisor activity from production use, so establish which one a flagged entry reflects before conceding a finding based on it.
Does the option apply on Standard Edition or in the cloud?
It exists only on Enterprise Edition, so a compressed attribute arriving in a Standard Edition database during a downgrade is a configuration error to fix rather than a licence to buy.
In cloud, some service tiers bundle the option while a bring your own licence move carries the on premises entitlement question with it, so confirm which applies before paying twice.
When is the exposure usually created?
During migration projects, which are the usual patient zero. Attributes move unseen with the data, the target estate goes live, and the project team disbands before anyone sweeps for feature usage.
A licence verification run in the week after go live is the cheapest control available and almost nobody schedules it.
How to Prepare for Your Oracle SaaS Negotiation
The 90-day renewal proposal with a 9 to 12 percent uplift is the bill for not preparing. The ARR compensation game, the utilization audit that finds 30 to 50 percent shelfware, benchmarks targeting 0 to 3 percent, one costed alternative, and sequencing toward May 31.
