Most priced options are already installed, several meter themselves from routine work, and each one drags a prerequisite behind it. Here is the arithmetic before Oracle sends it.
Enterprise Edition ships with most of its priced options already linked into the binary, and several of them record usage from work nobody thinks of as a decision. This page prices that risk: what is installed by default, what switches the meter on, which options drag others in behind them, and what the bill looks like when Oracle finds it.
Enterprise Edition is the base product. The options and management packs sit on top, each one a separate purchase, and buying the database entitles you to none of them.
That separation is where the money hides. A feature can be one click or one parameter away, and that click can create a six figure gap on a single server.
Most of them. A standard Enterprise Edition installation links the majority of priced options into the binary at install time, whether you bought them or not.
Because the options are linked in, they are available to any administrator the moment the database opens. Nobody has to install anything, request anything, or approve anything.
This is why the interpretation question matters so much. Presence in the option view and recorded usage in the feature usage view are entirely different claims, and our column by column reading guide takes that distinction apart properly.
Oracle ships a utility in the database home that enables or disables specific options in the binary itself. On 19c it covers Partitioning, OLAP, Advanced Analytics, and Real Application Testing.
Disabling an option you have not bought is a legitimate, dated configuration change that stops future usage at the source. Run it in a maintenance window with the database down, record the change, and keep the output.
More than most DBAs realize. Six categories account for nearly every accidental finding we see, and none of them involve a project decision.
The automated maintenance window includes an automatic SQL tuning task that invokes SQL Tuning Advisor. That advisor is a Tuning Pack feature, and the task is enabled by default.
Alongside it, the automatic workload repository takes snapshots on a schedule, which is Diagnostics Pack usage. Both are governed by CONTROL_MANAGEMENT_PACK_ACCESS, which defaults to enabling both packs.
The action, the option it meters, and whether it was a decision
| Routine action | Option or pack it meters | Deliberate? |
|---|---|---|
| Nightly automatic SQL tuning task | Tuning Pack, then Diagnostics Pack | No, on by default |
| AWR snapshot or ADDM report | Diagnostics Pack | No, on by default |
| Compressed backup set or export | Advanced Compression | Usually a DBA convenience |
| Standby opened for reads during apply | Active Data Guard | Often a reporting request |
| INMEMORY_SIZE set above zero | Database In Memory | One parameter, large bill |
| Fourth pluggable database created | Multitenant | Almost never noticed |
| Partitioned dictionary objects | Partitioning, as a false positive | No, Oracle managed |
White Paper · Oracle Database
Oracle Options & Management Packs
Why the options cost more than the database. Read it free.
Four chains do most of the damage, and none of them are obvious from the price list. Each one converts a single flag into a multiple line claim.
Tuning Pack is not sold standalone. If an advisor run puts Tuning Pack in the finding, Diagnostics Pack comes with it, and the combined list price is roughly 12,500 dollars per processor.
An option must be licensed on every processor where the database it runs on is licensed, at the same metric and the same quantity. There is no partial coverage and no per feature carve out.
This is the single most expensive rule on the page. One partitioned table on a 16 processor cluster is a 16 processor Partitioning liability, not a one table one.
You cannot license the database by Named User Plus and its options by processor, or the reverse. Named User Plus also carries a minimum of 25 users per processor on Enterprise Edition, which is set out in the Oracle Database licensing information manual.
Enough that the arithmetic, not the argument, is usually what ends the conversation. The numbers below are Oracle list prices per processor and should be verified against the current price list before you use them in a business case.
Indicative Oracle list price per processor, and the usual reason it appears
| Option or pack | List per processor | Named User Plus | Usual reason it is found |
|---|---|---|---|
| Enterprise Edition | 47,500 | 950 | The base the options attach to |
| Partitioning | 11,500 | 230 | Performance work, or dictionary false positive |
| Diagnostics Pack | 7,500 | 150 | AWR on by default |
| Tuning Pack | 5,000 | 100 | Automatic tuning task, needs Diagnostics too |
| Advanced Compression | 11,500 | 230 | Backup, export, or secure file compression |
| Advanced Security | 15,000 | 300 | Transparent data encryption or redaction |
| Active Data Guard | 11,500 | 230 | Reporting on an open standby |
| Real Application Clusters | 23,000 | 460 | Clustered availability, every node |
| Database In Memory | 23,000 | 460 | One parameter set above zero |
| Multitenant | 17,500 | 350 | Container count past the free allowance |
| Real Application Testing | 11,500 | 230 | Upgrade and replay tooling |
Multiply physical cores by the Oracle core factor to get processor licenses, then multiply by the list price of each option. Most modern Intel and AMD cores carry a factor of 0.5 in the Oracle processor core factor table.
Our page on the Oracle core factor works through the edge cases. The rest of this section assumes you already have the processor number right.
Take one two socket server with 16 cores per socket. That is 32 cores, a 0.5 core factor, and therefore 16 processor licenses for every option running on it.
That is one server, three findings, none of which anyone chose. Multiply it across a cluster and you have the opening claim most buyers see.
Source: Redress Compliance advisory engagement file, 2024 to 2025.
An Oracle option is not a feature you own because you bought the database. It is a separate meter, and on most estates the meter was switched on by the installer, not by a person.
By closing access, not by deleting history. Every legitimate lever on this page prevents future usage, and none of them touch the record of what has already happened.
Set the management pack access parameter to NONE on every database that does not carry a Diagnostics or Tuning entitlement. Set it to DIAGNOSTIC where you own Diagnostics but not Tuning.
Do it as a ticketed change with a date, an approver, and a before and after value recorded. That record is what turns a configuration decision into evidence.
Pull the feature usage and high water mark views across every Enterprise Edition instance, and store the result with a date. You are reading exactly the data Oracle will read.
Once the file exists, the interpretation work belongs in the reading guide, and the discipline for handling an Oracle supplied script belongs in the collection run guide.
Lay the usage list beside your contract entitlements and split it three ways. The gap between used and owned is exposure, the gap between owned and used is waste, and both are worth money at a renewal.
Five, in this order. The order matters more than the individual actions, because each one narrows what the next has to cover.
Set pack access correctly, disable the automatic tuning task where Tuning Pack is not owned, and relink out the options you have no intention of buying.
Require a license check before any administrator enables a paid option, and put the list price in the change template. Cost visible at the moment of the decision changes the decision.
Standard Edition 2 carries no separately licensed options, so the entire option exposure disappears for workloads that fit. The constraints are real, including the socket ceiling and the loss of clustering from 19c, so test before you commit.
Drift is normal and quarterly sweeps catch it while it is still a configuration problem rather than a finding. Diff each sweep against the last and investigate every new row.
An option gap raised at renewal is a commercial negotiation with something to trade. The same gap raised inside an open audit is a compliance claim with backdated support attached, and the sequencing work belongs in the pre renewal footprint program.
The standard reseller line is that you should license every option you might ever touch, to stay safe. We disagree. In roughly seven out of ten estates we baseline, the buyer is paying for or being chased on options no production workload actually needs. The buyer side move is the reverse: measure real usage first, close the defaults that meter without a decision, isolate the few options a workload genuinely depends on, and license only those. Breadth is not safety. Breadth is a larger permanent support base that the next audit will still try to grow.
No. Each option and management pack is a separate purchase on top of Enterprise Edition. What confuses buyers is that most options are linked into the binary at install time, so they are physically present and usable without ever having been bought.
Diagnostics Pack and Tuning Pack, by a wide margin. Both are reachable by default and the automated maintenance tasks exercise them without human involvement, so the database records usage on estates where nobody believes anyone used them.
No. Tuning Pack requires Diagnostics Pack as a prerequisite, so a single advisor run creates two line items rather than one. At list that is roughly 12,500 dollars per processor before support.
Yes, on every processor where the underlying database is licensed, at the same metric and quantity. One partitioned table on a 16 processor cluster is a 16 processor Partitioning liability, which is why single object findings produce large numbers.
Setting it above zero enables the option and records usage. Recent 19c releases include a base level of up to 16 gigabytes that does not require the full option, but it must be configured with the matching force parameter to stay inside the allowance.
Not since 12c. Native network encryption is included with all editions, so an opening claim that prices Advanced Security for encrypted SQL traffic is wrong on its face. Transparent data encryption and data redaction are the features that do require the option.
Yes, and it is legitimate. The relink utility in the database home disables specific options in the binary, which stops future usage. It does not remove historical usage rows, and you should never delete those.
For workloads that fit, yes, because Standard Edition 2 has no separately licensed options. The constraints are genuine, including a socket ceiling and the removal of Real Application Clusters support from 19c, so validate performance and availability before committing a workload.
Every option and pack licenses on the full processor count of the database beneath it, and the two cheapest switch on by default. The worked math and the strip and prove playbook.
Used across more than five hundred enterprise engagements. Independent. Buyer side. Built for procurement leaders running the next renewal cycle.
Oracle options are not features. They are meters. The estates that win the audit are the ones that read their own meters before Oracle reads them.
