JD Edwards concurrent licensing. A peak you have to prove.

A concurrent license is a promise about a moment, not about a population. It says that at no instant will more than a stated number of sessions be live. Nobody in your estate is measuring that moment, which is the whole problem.

White Paper · Oracle JD Edwards

Oracle JD Edwards Licensing

Keep JD Edwards licensing under control. Read it free.

Read the white paper

What does a JD Edwards concurrent license actually count?

It counts the largest number of qualifying sessions that were live at the same instant inside a defined measurement period. Three variables sit inside that sentence, and your contract sets all three: what qualifies as a session, how long the period is, and what counts as the same instant.

Get any one of the three wrong and the number changes materially. Oracle documents the product itself on its JD Edwards EnterpriseOne page, but the metric wording lives only in your ordering document.

What counts as one session?

• A signed in human at one workstation. Uncontroversial, and rarely the issue.
• The same human in three browser tabs. Depending on the wording, that can be one connection or three, and the difference across a thousand staff is not small.
• A process holding an application connection. Batch, interface and monitoring identities hold connections the same way a person does, and a log cannot tell them apart without help.
• A connection left open by a closed browser. Sessions do not always end when the human walks away. They end when the timeout says they do.

What period is the peak measured over?

Whatever period the contract names, and if it names none, whatever period the party with the data proposes. A month containing a period close, a stock count and a payroll run will produce a different peak from a quiet fortnight in the same year.

Insist on the period being stated before any data is produced. Producing data first and arguing about the period afterwards is a losing sequence, and it is the sequence most estates fall into.

Which words in your order decide the number?

Four phrases carry the weight. Find them in your paper before you accept anybody's arithmetic, including your own.

1. The unit. Concurrent user, concurrent device and concurrent session are three different units with three different counts.
2. The qualifier. Some wordings count only sessions performing a transaction. Others count any active connection.
3. The scope. Whether the quantity applies per component, per instance or across the enterprise changes everything downstream.
4. The measurement clause. If the order says how measurement will be performed, that method binds both sides. If it is silent, method is negotiable and you should propose one first.

Can you still buy concurrent licensing for JD Edwards?

Not from the current component price list, in our reading of it. Concurrent quantities in live estates arrived on older paper, and Oracle's published price lists moved to the application user and employee metrics long ago. Check the current Oracle price lists and your own order rather than taking anybody's word for it.

Where do surviving concurrent quantities come from?

• Pre acquisition agreements. Contracts signed with J.D. Edwards or with PeopleSoft before Oracle acquired the line, carried forward through assignment.
• Early Oracle era orders that preserved the incumbent metric to close a deal quickly.
• World customers whose paper reflects a different product generation and a different counting tradition entirely.

Legacy paper is usually an asset. It is also fragile, because it depends on documents that are thirty years old in some estates and on definitions Oracle has no commercial interest in reaffirming.

What happens when you buy something new?

A new purchase is the moment the legacy metric comes under pressure. Oracle will usually propose that the estate standardizes on current metrics, and the concurrent quantity gets converted into an application user quantity at a ratio.

There is no published ratio. It is derived from evidence, and if the only evidence in the room is a list of named accounts, that list becomes the ratio. This is why a maintained concurrency baseline is worth more than any clause you could negotiate about it later.

How is a concurrent peak actually measured?

It is reconstructed, never read off a dial. JD Edwards has no single counter that produces a defensible concurrency figure, so somebody assembles one from several systems, and the assembly method decides the answer as much as the estate does.

Which systems hold the evidence?

Read the retention column again. It is the single most consequential line in this article, because an audit request will reach back further than your default log settings survive.

How does the sampling method change the answer?

A peak is only meaningful once you say how often you looked. Sampling every sixty seconds catches spikes that a five minute sample never sees, and an estate with bursty transaction patterns can differ by a fifth between the two methods.

• Instantaneous peak. The highest single reading anywhere in the period. Volatile, and the number an auditor prefers.
• Sustained peak. The highest reading held for a stated duration, such as five consecutive minutes. Far closer to what the business actually needs.
• Percentile. The ninety fifth or ninety ninth percentile of readings. Statistically honest, and worth proposing where the contract is silent on method.
• Business day filter. Whether overnight batch windows are inside or outside the measured period is a decision, not a fact.

What happens if you have no concurrency data at all?

Then the conversation stops being about concurrency. In practice, the fallback position becomes the number of authorized accounts, because that is the only figure both parties can see, and it is the highest figure available.

That fallback is not a contractual rule. It is what happens when one side has data and the other has an assertion. Oracle License Management Services is not obliged to accept an estimate you cannot reproduce.

When does concurrent beat the application user metric?

When the ratio of authorized individuals to defensible peak is larger than the price multiple between the two metrics. That is the whole test, and it is arithmetic rather than judgement once you have both numbers.

A worked break even for a three shift manufacturer

Now put your own price list against it. If a concurrent unit costs less than six times an application user for the same component, concurrent wins at this ratio. If it costs more, the legacy metric is sentiment rather than strategy.

Where the common advice on JD Edwards concurrent licensing is wrong

The common advice is to defend the concurrent metric at all costs, because it is almost always cheaper than named licensing. We think that advice is right about the price and wrong about the risk.

A metric you cannot measure is a liability, not a saving. If your logs rotate in a fortnight, your sign on auditing has never been switched on, and your paper does not define a measurement method, then your concurrent quantity is an unpriced option that Oracle can call at a moment of its choosing.

The sequence that actually protects value is unfashionable, and it is only three steps.

1. Build the concurrency baseline first, and keep it for years rather than for weeks.
2. Decide deliberately whether you intend to defend the metric or to trade it.
3. If you trade it, do so at a moment when you are buying something Oracle wants to sell.

Converting from strength costs less than defending from ignorance. The second route is the one most estates take by default, because nobody chose the first one in time.

Concurrency is not a property of your estate. It is a property of your evidence. Change the evidence and you change the license position, entirely lawfully.

What does Oracle ask for when concurrent quantities are in scope?

Something broader than concurrency, almost always. The request that lands is usually a standard applications data collection, and concurrency is not what it is designed to produce, which puts you in the position of volunteering the only evidence that helps you.

The five things the request usually reaches for

1. A full extract of user profiles and security records, which produces a headcount, not a peak.
2. Role and menu assignments mapped to components, which produces authorization, not simultaneity.
3. Environment and instance inventory, including the environments people forget: training, disaster recovery, and the copy somebody built for a migration test.
4. Installed component evidence, which sets the scope of everything that follows.
5. Any measurement output you happen to hold, which is the only line in the request where your baseline can enter the record.

Read that list again from a negotiating position. Four of the five items build the case for a named user count, and the fifth is optional unless you supply it.

The sequence that protects the metric

• Answer with your own measurement first. Lead the pack with a dated, method described concurrency series rather than burying it behind a user extract.
• State the method before the number. Sampling interval, period, inclusion rules and exclusions, all written down before the figure appears.
• Explain every exclusion with its identity list. A removed machine session that cannot be named is an argument. One that comes with an interface name and an owner is a fact.
• Keep the non production environments in a separate schedule, with their access populations and their purpose, so that a test copy never gets folded into a production peak.

Does any of this change for JD Edwards World?

The commercial logic is the same and the evidence is different. World estates sit on a different technology generation, so the session evidence comes from the platform layer rather than from a web tier, and the contract vintage is usually older still.

Confirm which product line your order actually names before you spend a week extracting the wrong data. An order that names one product line does not license the other, and estates running both need to keep the two counts strictly apart.

What moves cut a JD Edwards concurrent count?

Six moves do real work, and five of them are configuration rather than negotiation. Do them in this order, because each one makes the next one cheaper to prove.

The configuration moves

• Set an idle timeout that matches a written policy. An arbitrary timeout is a setting. A timeout that implements a documented policy is evidence.
• Give machines their own identities. Interfaces, monitors and scheduled jobs should never share an identity with a human population you are trying to count.
• Constrain parallel connections where the platform allows it, so that one person cannot present as three.
• Close sessions on sign out properly, including the kiosk and shared workstation pattern where staff simply walk away.
• Sample continuously and store the samples. A small time series of session counts, retained for years, costs almost nothing and is the artefact that wins the argument.

The one commercial move

Decide, in advance and in writing, whether you intend to defend the concurrent metric or to trade it. Estates that never make that decision end up converting under time pressure during a purchase, which is the worst available moment.

If you intend to trade it, bank the concurrency baseline first, then use the conversion as consideration for something you actually want: a component swap, a termination right, or audit language with teeth. Read our note on JD Edwards user metrics before you agree any ratio, because the metric you convert into carries its own counting rules.

What should a buyer do next?

1. Find the original ordering document that created the concurrent quantity, and read the metric wording out loud to the application owner.
2. Establish today whether sign on auditing is enabled and how long each log source is retained, before anybody asks you for history.
3. Stand up a lightweight session sampler that records a timestamped count every minute and writes it somewhere with multi year retention.
4. Separate machine identities from human ones so that future samples are already clean.
5. Produce a first baseline over a representative period, with the method written down and dated.
6. Calculate your ratio of authorized individuals to defensible peak, and test it against the price multiple in your own price list.
7. Decide deliberately whether to defend or convert, and engage independent Oracle advisory before the next order, not after it.

Frequently asked questions

What does JD Edwards concurrent licensing count?

It counts the largest number of qualifying sessions live at the same instant within a defined measurement period. The definition of a qualifying session, the length of the period and the sampling method all come from your ordering document rather than from the software.

Can I still buy concurrent licenses for JD Edwards?

Not from the current component price list, in our reading of it. Concurrent quantities in live estates came in on older agreements, often pre acquisition paper, and Oracle now sells the application user and employee metrics instead. Verify against your own order and the current price list.

How does an auditor measure a concurrent peak?

By reconstructing it from logs, because JD Edwards does not ship an audit grade concurrency counter. Web tier session records, proxy logs, sign on auditing and batch job records are assembled into a time series, and the assembly method matters as much as the estate does.

What happens if I cannot produce concurrency evidence?

The discussion defaults to counting authorized accounts, which is almost always the most expensive available number. This is not a contractual rule, it is simply what happens when one party holds data and the other holds an opinion.

Do batch jobs and interfaces count toward the concurrent peak?

They appear in the raw readings, so they count unless you can separate them. Give machine processes their own identities well before any measurement, because retrospectively distinguishing them inside a shared log is slow, expensive and easy to challenge.

How long should I keep session data?

Longer than your audit lookback, which usually means years rather than the days or weeks that default log rotation provides. A one line per minute session count is tiny to store and is the cheapest insurance in the whole licensing estate.

How is a concurrent quantity converted to application users?

By negotiation, from whatever evidence exists at the time. There is no published conversion ratio, so the ratio ends up anchored on the best documented number in the room, which is a list of named accounts unless you have built something better.

Is concurrent always cheaper than named user licensing?

No. Concurrent wins only when the ratio of authorized individuals to defensible peak exceeds the price multiple between the two metrics for that component. Below that break even it is more expensive, and it carries measurement risk that named licensing does not.