A buyer side guide to Oracle GoldenGate licensing for non Oracle databases and Big Data in 2026. Why both ends of the path count and where cores get missed.
Oracle GoldenGate for non Oracle databases and Big Data is licensed per processor on both ends of the replication path, so the cost question is the core count on every source and target, not just the Oracle system.
This guide is for data, integration, and procurement leaders running GoldenGate across mixed platforms in 2026. Pair it with the GoldenGate licensing guide and the Oracle Practice so the architecture and the entitlements match.
GoldenGate licenses per processor core on the systems running its processes. There is an edition for Oracle databases, a distinct edition for non Oracle databases, and a separate product for Big Data targets.
Oracle describes the product family on its Oracle GoldenGate page. The licensing detail to confirm is which edition applies to each platform you run.
The Oracle edition covers Oracle Database sources and targets. The non Oracle edition covers databases such as SQL Server, MySQL, and PostgreSQL. GoldenGate for Big Data covers streaming and analytics targets.
GoldenGate for Big Data delivers change data into streaming and analytics platforms rather than into a relational target. It licenses per processor on the delivery systems, so the Big Data side carries its own core count.
You count where GoldenGate runs its capture and delivery processes. That is usually both the source and the target. The target side is the part most estates forget when they budget.
GoldenGate licensing footprint across a replication path (confirm per current terms)
| Point in the path | Process | Licensed? |
|---|---|---|
| Source database | Capture (extract) | Yes, per processor |
| Target database | Delivery (replicat) | Yes, per processor |
| Big Data target | Delivery to stream or store | Yes, Big Data edition |
| Pass through hub | Routing only | Depends on processing |
Budgets tend to start from the Oracle source and stop there. The target database or Big Data platform runs a delivery process that also needs licensing, and that omission surfaces in an audit.
A path that crosses Oracle, a non Oracle database, and a Big Data target uses three different licensing treatments at once. Counting the whole path correctly means tracking each platform's edition and cores separately.
Consolidate capture and delivery onto fewer well sized cores, confirm the right edition per platform, and retire replication paths no longer in use. The core count on both ends is the lever.
Paths built for a migration that has finished, or for a report feed now served another way, often keep running and keep consuming licenses. Retiring them removes cost with no operational loss.
GoldenGate licenses per processor core on the database it runs against, and a separate edition covers non Oracle databases. The metric to confirm is the core count on the source and target systems, because GoldenGate is licensed where it captures and where it delivers, not once.
No. Oracle sells a distinct GoldenGate edition for non Oracle databases, and the rate and terms differ from the Oracle Database edition. Assuming one price across both is a common budgeting error, so price the non Oracle edition on its own terms.
GoldenGate for Big Data is a separate product that delivers change data into targets such as Kafka, Hadoop, and cloud object stores. It licenses per processor on the systems running the delivery process, so the Big Data targets, not just the source, drive the count.
Generally yes. GoldenGate captures changes on the source and applies them on the target, and both sides run licensable processes. The frequent surprise in an audit is target side cores that were never counted at purchase.
Yes, by consolidating capture and delivery onto fewer, well sized cores, by confirming the correct edition for each platform, and by retiring replication paths no longer in use. The core count on both ends is the lever, so shrinking it shrinks the bill.
The biggest risk is uncounted target side cores and the wrong edition applied to a non Oracle or Big Data platform. Because GoldenGate spans heterogeneous systems, the estate is easy to under count when only the Oracle source is in view.
Oracle ULA exit moves, Java audit defense posture, certification framework, and the buyer side moves across the Oracle Database, Java, and EBS estate.
Used across more than five hundred enterprise engagements. Independent. Buyer side. Built for procurement leaders running the next renewal cycle.
Budgets start at the Oracle source and stop there. The delivery side, a non Oracle or Big Data target, carries license cost that surfaces only in an audit.
500+ enterprise clients. 11 vendor practices. Industry recognized. One conversation can change what you pay for the next three years.
One short note on Oracle data integration, GoldenGate, and the buyer side moves we are running in client engagements.