Legacy Named User carries a 10-per-processor minimum against today's 25 NUP floor, so most baselines under-claim entitlement by roughly 60 percent on those lines
Named User (2001 re-issue), Universal Power Unit, and Concurrent Device entitlements are still live perpetual assets on most Oracle estates, and almost every license position values them wrong in both directions. The 10-per-processor minimum on legacy Named User is worth 15 processors of coverage where 6 would be claimed under Named User Plus, while UPU on modern hardware can demand 10 to 20 times the units the same workload needed in 2001. Get these three lines right before you accept any audit finding or renewal quote.
Prepared by Redress Compliance · August 20, 2026 · Oracle advisory. Legacy entitlement reviews and audit defense engagements, 2024 to 2026.
Executive summary
The 2001 Named User re-issue carries a 10-per-processor minimum for Database Enterprise Edition against the 25-per-processor minimum on current Named User Plus, a 2.5x advantage most baselines throw away.
A 150 Named User entitlement legitimately covers 15 Enterprise Edition processors; the same 150 counted as NUP covers 6, and the 9 processor gap gets quoted back to you at $47,500 list each.
Concurrent Device licenses were sold with no per-processor minimum, yet Oracle's own Oracle8i Pocket Sales Guide asserts 5 concurrent devices per processor and LMS routinely applies it to contracts that contain no such term.
House of Brick has documented Oracle sales and LMS attempting to impose per-processor restrictions on customers whose paper is silent, which means the defense is contractual, not technical: produce the order form and the assertion collapses.
Universal Power Unit is the opposite risk, because the same UPU count that licensed a 2001 server can require 10 to 20 times more units on a modern multi-core box, and the widely cited 1,000 UPU to 1 Processor conversion has no contractual standing.
Oracle stopped selling UPU in 2001, so any migration to Processor licensing is a negotiated event where you should anchor on 1,000:1 before Oracle anchors somewhere worse.
Every valuation error compounds at 2.1x because support runs at 22 percent of net license fees, so one dollar of mis-valued legacy entitlement costs roughly $2.10 across a five-year hold.
On Enterprise Edition that is $10,450 of annual support per processor you either avoid or absorb, which is why the legacy lines deserve the same forensic attention as the current-metric estate.
How the three legacy metrics actually count, and where the counting breaks
Three metrics sit on most Oracle estates that predate the current price list, and each one counts on a basis your SAM tool does not model. Named User exists in at least three shapes: Named User Single Server (NU-SS), tied to a user against one specific machine.
Named User Multi Server (NU-MS), permitting a user to reach multiple servers.
And the 2001 re-issue, which carries a minimum of 10 Named Users per processor on Database Enterprise Edition and 5 or 10 on Standard Edition depending on purchase date. Universal Power Unit (UPU, sometimes written UPP) arrived in 2000, replacing the 1999 Power Unit Intel and RISC metrics.
And calculates entitlement from processor clock speed and architecture type.
Oracle stopped selling UPU in 2001 when the Processor metric arrived, and Processor is still on the list today at $47,500 per processor for Enterprise Edition. Concurrent Device counts the maximum number of input devices accessing a program at any given point in time.
Measured at the multiplexing front end where multiplexing hardware or software sits in the path.
The end-of-sale date for Concurrent Device is genuinely contested across advisory sources, with several dating withdrawal to around 1999 and others later, which means you check your own order forms rather than a published timeline.
In every case the counting basis on the paper governs, and the direction of error is predictable per metric, as the table sets out.
| Metric | Sale window | Counting basis | Per-processor minimum | Usual direction of baseline error |
|---|---|---|---|---|
| Named User (2001 re-issue) | 2001 onward, superseded by NUP | Authorized individuals | 10 per processor (EE); 5 or 10 (SE) | Under-claimed: tool applies the 25 NUP floor |
| Named User Single Server (NU-SS) | Pre-2001 | Individuals against one named server | 30 UPU per Named User (EE) | Mis-scoped: multi-server use unlicensed |
| Named User Multi Server (NU-MS) | Pre-2001 | Individuals across multiple servers | 30 UPU per Named User (EE) | Under-claimed: multi-server right ignored |
| Universal Power Unit (UPU/UPP) | 2000 to 2001 | Clock speed times architecture factor | Not applicable | Over-exposed: modern cores need 10 to 20x |
| Concurrent Device | 1990s, withdrawn approximately 1999 | Peak simultaneous input devices | Not applicable | Over-exposed: counted as named individuals |
The single highest-value find in a legacy stack is not a metric rule at all, it is a piece of paper. Named User licenses purchased through a Network License Order Form, common in that period, frequently carry no minimum whatsoever.
Where no minimum is referenced in the order document, the customer may run the programs across any number of servers provided the Named User count itself is not exceeded. No 10-per-processor floor, no 25-per-processor floor, no processor count in the equation at all.
That exception cannot be expressed as a row in the table above because it is document-dependent, not metric-dependent: two customers holding identically labeled Named User licenses can have completely different obligations depending on which order form was used.
In a VMware estate the difference is decisive, because Oracle's standard position is to demand licensing of every processor in the cluster or the vCenter. A no-minimum Named User order form removes the processor from the argument entirely.
Pull the original order forms before you accept any deployment-based finding, as covered in reconciling Oracle entitlements to deployment.
Named User Legacy: the under-claimed asset and the batching trap that offsets it
The 2001 Named User re-issue is under-claimed on almost every position we review, and the mechanism is mechanical rather than negligent.
SAM tools normalize entitlement lines into current metrics for comparability, so a Named User line becomes a Named User Plus line, and the moment it does the tool applies the current 25-per-processor floor instead of the contractual 10.
Work the arithmetic on 150 Named User licenses against Enterprise Edition. At the legacy 10-per-processor minimum those 150 users support 15 processors of deployment. Normalized to NUP at 25 per processor, the same 150 users support 6 processors.
That is a 9 processor gap, and at the April 2026 list price of $47,500 per Enterprise Edition processor the under-claim is worth $427,500 at list, before the option stack. Add Partitioning at $11,500 and RAC at $23,000 per processor and the same nine processors carry another $310,500.
Support compounds it: at 22 percent of net license fees, every dollar of understated entitlement you buy back unnecessarily costs roughly $2.10 across a five year hold once the recurring support line is counted.
The offsetting exposure sits in a feature the legacy metric never included. Current Named User Plus permits automatic batching between relational databases, meaning users on a non-Oracle front end whose data reaches the Oracle database only via scheduled batch require no Oracle license.
Legacy Named User does not include that relief.
In our experience Oracle's audit teams know this and probe for it specifically, because it converts a whole population of upstream users into countable Named Users on the legacy lines while the same population is exempt on the NUP lines sitting one row below.
Scope compounds the risk: NU-SS ties a user to one named server, so a user reaching a second instance is unlicensed on that instance, while NU-MS carries the multi-server right and is frequently the more valuable line once found.
On the pre-2001 variants a second constraint applies, the 30 UPU per Named User minimum on Enterprise Edition, which means a dual-CPU 1.5 GHz Xeon consuming 3,000 UPUs implies 100 NU-SS or NU-MS licenses regardless of how few humans actually connect.
Net position: claim the 10-per-processor minimum in writing, quote the order form clause, and simultaneously scope your batch-fed user populations before Oracle does. Both corrections belong in the same submission so the arithmetic reads as accuracy rather than opportunism.
The counting rules on the current metric are set out in Oracle Named User Plus counting rules and per-processor minimums.
Named User Plus or Processor: The Metric Decision
Oracle sells the same database per user and per processor. The 25 NUP per processor minimum, the 50 user break-even, and how to choose the metric that fits the workload.
Get the white paper →Why legacy metrics are worth more in a negotiation than in a compliance spreadsheet
A legacy entitlement line is not really a license count. It is an undated call option written against Oracle's own historical pricing logic, and options are worth more before they are exercised than after.
When you drop a Named User (2001) line into a compliance reconciliation, you are exercising it: you convert an ambiguous, arguable, contractually anchored quantity into a single number that either matches deployment or does not. That conversion destroys the thing that made the line valuable.
The 10-per-processor minimum, the Network License Order Form with no minimum referenced, the Concurrent Device line with no per-processor floor at all, these are only worth something while they remain unconverted and unpriced.
The moment they enter a spreadsheet as "equivalent to X NUP," you have accepted Oracle's translation of your paper into Oracle's current commercial vocabulary, and you did it for free.
Understand why Oracle wants these lines gone. They are unmeasurable by Oracle's own tooling.
LMS scripts and the standard review outputs count processors, cores, and users against current definitions; nothing in that kit computes UPU against clock speed and architecture, and nothing determines a concurrent device count at the multiplexing front end.
They are also unpriceable on the current price list, because UPU new sales stopped in 2001 and Concurrent Device disappeared around 1999. There is no list anchor, no support ratio to extend, no discount schedule that maps cleanly.
For an audit team whose entire method is "measure with our tool, price against our list," a legacy line is a hole in the process. That hole is your leverage, and it is the only thing on the table Oracle cannot resolve unilaterally.
This is exactly why audit teams behave the way they do on these lines. Pushing a per-processor minimum onto Concurrent Device, where the original paper carries none, is not a contract interpretation, it is a normalization move designed to make an unpriceable line priceable.
Presenting the 1,000:1 UPU-to-Processor ratio as settled fact is the same move: House of Brick's own reading is that the ratio is not formally binding and must be negotiated, yet it arrives in findings as arithmetic rather than as a proposal.
Once you accept the ratio, the negotiation over the ratio is over, and everything after that is a discussion about volume. Treat every normalization as a commercial ask in technical clothing.
The asymmetry here runs in the buyer's favor, and most license managers never use it. Your obligation is to produce the order form. Oracle's obligation, if it wants a restriction that is not on that order form, is to prove where the restriction comes from.
A minimum that appears in a 2026 price list does not retroactively attach to a 2001 ordering document. A conversion ratio published in nobody's contract is not a term.
That evidentiary burden is why the paper trail matters more than the tooling output, and why any buyer-side baseline built before a renewal or audit should record legacy lines with their originating document reference rather than with a modern equivalent.
If you cannot cite the order form, you have already conceded the argument.
The strategy follows directly. Hold the legacy lines unconverted through the compliance discussion. Settle the measurable estate, the current Processor and NUP lines, in the reconciliation, and carry the legacy lines forward as an unresolved item.
Then price the conversion yourself, at a renewal, when Oracle wants something from you.
On Enterprise Edition at $47,500 per Processor and $950 per NUP against a 40 to 70 percent enterprise discount benchmark, you are converting into a market where the real unit is between $14,250 and $28,500, not list.
Every legacy processor-equivalent you hand back at list is roughly two to three times the price you would have paid for the same coverage in a negotiated deal.
Add the support multiplier and the case closes. Support runs at 22 percent of net license fees, so a dollar of net license value is worth roughly $2.10 over a five-year hold.
A conversion negotiated at 60 percent off, rather than accepted at list inside an audit settlement, does not save you the discount once, it saves it every year the support stream runs.
In my experience across legacy entitlement work, buyers lose more money by converting cleanly and early than by ever being found out of compliance on a metric Oracle cannot measure.
UPU and Concurrent Devices: the two lines that move in opposite directions
These two metrics sit adjacent in the same 1999 to 2001 vintage of paper, and they fail in opposite directions. UPU is hardware-derived, so it inflates every time you refresh. Concurrent Device is user-derived and carries no per-processor floor, so it stays flat while your server count grows.
UPU was computed from processor clock speed and architecture type, a formula that made sense when a dual-socket 1.5 GHz Xeon was a large box.
Run the same formula against a modern multi-core server and the requirement rises by 10 to 20 times for identical workload, with RISC platforms (IBM pSeries, HP PA-RISC, Itanium, SPARC) carrying a formula penalty that pushes the result higher again than x86 of comparable vintage.
Because Oracle stopped new UPU sales in 2001, there is no current list price to true up against, which cuts both ways: Oracle cannot quote a UPU shortfall from a price list, and you cannot buy your way out of one either.
| Dimension | Universal Power Unit | Concurrent Devices |
|---|---|---|
| Basis of count | Processor clock speed and architecture | Maximum simultaneous input devices |
| Direction of drift | Over-exposed, grows with every refresh | Under-claimed, flat as servers multiply |
| Per-processor minimum | None, but hardware drives the number | None, the key efficiency |
| Measurement point | Physical CPU inventory | The multiplexing front end |
| Modern hardware effect | 10 to 20x the 2001 unit requirement | Neutral, device count is unchanged |
| End of sale | 2001, no successor list price | Contested, 1999 or 2001 depending on source |
| Conversion anchor | 1,000:1 to Processor, not contractually binding | No published ratio at all |
The two rows that decide the money are per-processor minimum and measurement point. Concurrent Device has no floor, which means a 500-device entitlement can legitimately cover an estate of twenty servers where the equivalent NUP position would carry a 25-per-processor minimum on every one of them.
That is the single most under-claimed line in the legacy stack, and it is worth checking against the current NUP minimum rules before you accept any restatement.
On UPU, the measurement point is your defense. The count is derived from hardware, so the correct response to a UPU finding is not to recount users but to challenge the formula applied and to refuse the 1,000:1 conversion as a settled term.
The end-of-sale date on Concurrent Device is genuinely disputed across advisory sources, 1999 in some readings and 2001 in others, which is precisely why your own signed paper, not any published timeline, governs what you hold.
What the paper trail shows across legacy entitlement reviews
A 10-per-processor legacy Named User minimum covers 15 processors on a 150-user line, while the same line mapped to the 25 NUP floor claims only 6, a 60 percent under-claim.
Modern multi-core servers frequently require 10 to 20 times the Universal Power Units the same workload consumed on 2001-vintage hardware, turning a dormant entitlement into an exposure.
Across legacy entitlement reviews the same four patterns recur, and none of them is accidental.
First, Oracle field teams assert minimums sourced from sales collateral rather than contracts: the Oracle8i Pocket Sales Guide five Concurrent Devices per processor rule gets quoted at customers whose order forms are silent on any minimum at all.
Second, LMS and the audit script output map every legacy Named User line to Named User Plus at 25 per processor, which is the single most expensive reading available and has no contractual basis unless your agreement was amended.
Third, SAM tools default legacy lines to zero (invisible entitlement) or silently normalize them to the current metric equivalent, which is why we tell clients to read the tooling comparison for building a license position before trusting any automated baseline.
Fourth, and most consequential, Network License Order Forms almost never live in the CLM repository.
In our engagement experience they surface in archive boxes, departed DBAs' filing cabinets, and acquired-entity due diligence folders, and they frequently carry no minimum at all, which permits unlimited server deployment against a fixed user count.
The source disagreement on Concurrent Device withdrawal is itself the evidence. House of Brick and SoftwareOne date the metric's retirement around 1999; other analysts place it later. That spread proves the point: no published date governs your rights, only your executed order form does.
Value each surviving line against the April 2026 anchors, $47,500 per Processor, $950 per NUP, $10,450 annual support per Processor, and the delta becomes quantifiable rather than rhetorical.
The asymmetry is the finding. Every recurring pattern above pushes the position in Oracle's favor: the sales guide minimum inflates the requirement, the 25 NUP mapping shrinks the entitlement, and the SAM default erases it entirely. Not one of the four defaults errs toward the buyer.
That is not tool immaturity, it is the predictable result of counting rules being interpreted by the party that profits from the interpretation.
Practically, this means the archive search is the highest-return hour in the entire baseline exercise. One Network License Order Form with no minimum clause is worth more in a VMware cluster dispute than six months of deployment discovery, because it removes the per-processor argument altogether.
Treat every legacy line as unvalued until you hold the paper.
- 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
Your first five moves
- Pull the physical archive, not the CLM system. Retrieve every pre-2002 order form, Network License Order Form, and acquisition-inherited schedule from storage, legal files, and departed DBAs' boxes, because contract management systems were populated from post-2002 migrations and routinely omit exactly the documents that carry the favorable minimums.
- Tag every legacy line by its own metric and minimum rule. Record whether each line is NU-SS, NU-MS, 2001 re-issue Named User, UPU, or Concurrent Device, with the minimum stated in that document (10 per processor, 30 UPUs per user, five CDs per processor, or none), and explicitly reject any tool-assigned current-metric equivalent until you have reconciled it to paper, following the discipline in the entitlements-to-deployment reconciliation approach.
- Build the dual valuation and price the delta. For each legacy line, model coverage at your contractual minimum and at Oracle's asserted equivalent, then convert the gap to dollars at $47,500 per Processor list plus $10,450 annual support, so a 9-processor swing on a single 150-user Named User line reads as roughly $427,500 of license and $94,050 of recurring support, and cross-check the counting rules against the NUP minimums and audit traps guidance.
- Refuse UPU to Processor conversion outside a renewal you control. The 1,000:1 ratio is a practice, not a contract term, so never accept a conversion inside an audit settlement or a mid-term true-up where you have no discount leverage; anchor at 1,000:1, insist the resulting Processor count is priced at your negotiated discount band (40 to 70 percent off list is the enterprise norm), and get the ratio written into the order document rather than an email.
- Bar any uncompensated surrender of legacy lines. Instruct procurement and the ULA certification team in writing that no legacy Named User, UPU, or Concurrent Device entitlement is retired, migrated, or folded into a new agreement without a priced conversion recorded in the executed order document, because these lines are perpetual assets and once they disappear from the entitlement schedule they cannot be reconstructed from support renewal invoices alone.
Frequently asked questions
Are legacy Oracle Named User licenses still valid if the metric is no longer on the price list?
Yes. Perpetual licenses granted under a Named User, UPU, or Concurrent Device metric remain valid under the terms of the order form and the governing agreement, and Oracle removing a metric from the price list does not retire your grant.
What Oracle can do is decline to sell you more of that metric, which is why the entitlement is a fixed-quantity asset. Keep the original order form, because the metric definition on your paper governs, not the current price list definition.
What is the per-processor minimum on legacy Oracle Named User compared with Named User Plus?
The Named User metric reintroduced around 2001 carried a minimum of 10 Named Users per processor for Database Enterprise Edition, and 5 or 10 for Standard Edition depending on purchase date. Current Named User Plus carries 25 per processor.
That 2.5x difference means a legacy Named User entitlement covers substantially more processors than the same count under NUP, which is why mapping legacy NU to NUP in a SAM tool systematically under-claims your position.
How many UPUs convert to one Oracle Processor license?
The commonly cited ratio is 1,000 Universal Power Units to 1 Processor license. It is important to understand this is not a contractual conversion right: it is a convention that must be negotiated with Oracle in each case.
Anchor on 1,000:1 in writing early, and never accept a conversion computed on today's core factor and clock speed logic, which can inflate the requirement by 10 to 20 times.
Can Oracle apply a 5 concurrent devices per processor minimum to my Concurrent Device licenses?
Only if your contract says so. Concurrent Device licenses were generally sold without per-processor minimums, although Oracle public documents such as the Oracle8i Database Server Pocket Sales Guide state a minimum license level of 5 concurrent devices per processor.
Oracle LMS and sales have been observed asserting that minimum against customers whose contracts contain no such term. Produce the order form and the assertion has no basis.
How do I count Concurrent Devices when multiplexing is in place?
The standard Concurrent Device definition is the maximum number of input devices accessing the program at any single point in time, and where multiplexing hardware or software sits in the path, the count is taken at the multiplexing front end.
That means the connection pool or middle tier does not reduce your count; you measure the devices in front of it. Model the peak, not the average, because the definition is a maximum.
Why is a Network License Order Form so valuable in an Oracle license position?
Named User licenses purchased on a Network License Order Form frequently carry no minimum at all, and where no minimums are referenced the programs may be used across any number of servers provided the Named User count is not exceeded.
That is exceptionally valuable in virtualized estates, where Oracle would otherwise demand licensing of every processor in a VMware cluster or vCenter. If one exists, it is usually the single highest-value document in the entitlement stack.
What is a legacy entitlement worth in cash terms if I get the count wrong?
Value it against current list and current support. Database Enterprise Edition lists at $47,500 per processor with $10,450 annual support, and Named User Plus at $950 with $209 support, on the 2026 Technology Global Price List.
Because support is 22 percent of net license fees, one dollar of valuation error costs roughly $2.10 over a five-year hold, so a nine-processor mis-count is not a $427,500 problem, it is closer to a $900,000 problem.
