Oracle to PostgreSQL Migration,
Planned Around Your PL/SQL
We move Oracle schemas, PL/SQL and data to PostgreSQL on the cloud you choose, with change-data-capture cutover so the business keeps running through it.
Want the engineering detail first? Read the step-by-step migration guide.
- ISO 27001 Aligned
- SOC 2 Type II Aligned
- US-based delivery director
What is Oracle to PostgreSQL migration?
Oracle to PostgreSQL migration moves an organization's databases off Oracle Database, a commercially licensed system, onto PostgreSQL, an open-source relational database with no licence fees. It covers the schema, the data, the PL/SQL code inside the database, and the SQL in the applications that use it. The code, not the data, is usually most of the work.
The target is normally a managed service: Amazon RDS for PostgreSQL, Amazon Aurora PostgreSQL, Azure Database for PostgreSQL flexible server, Cloud SQL for PostgreSQL or AlloyDB for PostgreSQL. Self-managed PostgreSQL stays the right answer where data residency or cost models point that way. Either way an Oracle to Postgres move is two projects running together: one inside the database, one inside every application that talks to it.
Why companies are moving from Oracle to PostgreSQL
Five reasons that come up in almost every business case we are asked to review, and what each one is actually worth.
Licence and support costs that grow with every core
Oracle Database Enterprise Edition is licensed per processor, with a core factor applied to the physical core count, and annual support is charged as a percentage of the licence fee. Options such as Partitioning, Advanced Security and Diagnostics Pack are licensed separately on top. PostgreSQL has no licence fee at all; you pay for infrastructure, managed-service charges and people. Set against that, a migration is a one-time cost that has to be paid back out of the saving, which is why the assessment comes before the business case rather than after it.
Less exposure to licence reviews and vendor lock-in
Licence compliance reviews are a recurring cost and risk for many Oracle customers, in audit preparation, internal time and the true-up that sometimes follows. The deeper issue is bargaining position: when the database cannot move, the renewal conversation has only one direction. PostgreSQL removes the licence position from that negotiation. It does not remove the work of running a database well, and anyone selling it that way is selling something else.
A free choice of cloud
PostgreSQL runs as a first-class managed service on AWS, Azure and Google Cloud, and self-managed on-premise. The database stops dictating the cloud decision, which matters most to organizations that have already decided to consolidate on a platform and found one system holding the plan up. It also means a second migration later is a smaller exercise than the first one.
A larger talent pool
In the 2025 Stack Overflow Developer Survey, PostgreSQL was the most used database among respondents who had done extensive development work with one in the past year, at 55.6%, ahead of MySQL at 40.5% and Microsoft SQL Server at 30.1%. Hiring for PostgreSQL is a different market from hiring for Oracle, and it moves in the opposite direction year on year. See the survey's database results.
Extensions instead of add-on licences
PostGIS for geospatial work, pgvector for the vector search behind retrieval-augmented AI features, JSONB for document-shaped data, and declarative partitioning in core. These are capabilities you add to a PostgreSQL instance rather than line items you license. The honest caveat is that they are not a feature-by-feature replacement for every Oracle option, and a few Oracle features have no PostgreSQL equivalent at all, which is the subject of the next section.
When moving off Oracle is the wrong call
Four situations where we have told clients not to migrate, or not to migrate that database yet.
Packaged applications certified only on Oracle
Oracle E-Business Suite and similar products decide their own supported database. Migrating underneath a vendor's application ends support for it, which is a larger problem than the licence.
Systems built around Oracle RAC or Exadata features
Where the architecture depends on RAC's shared-everything clustering or Exadata storage offload, the re-architecture cost can exceed the licence saving for years. That is a calculation, not a principle, and it is worth doing properly.
Large PL/SQL estates with little test coverage and a short remaining life
If the system is being retired in two years, converting tens of thousands of lines of undocumented PL/SQL buys very little. The test suite you would have to build first is most of the cost.
Cases where a better Oracle licence arrangement solves the real problem
Sometimes the complaint is a specific over-licensed environment, not Oracle itself. Consolidating or right-sizing that environment can resolve it without a migration programme.
The assessment tells you which of your databases fall into this group before you spend anything on migrating them. Deciding that per database rather than per estate is the core of a modernization strategy worth following.
What our Oracle to PostgreSQL migration services cover
Ten deliverables, from the first object inventory through to the weeks after cutover.
Assessment and object inventory
Tables, views, packages, procedures, triggers, sequences, types, synonyms, DB links, scheduler jobs
Target architecture and sizing
Managed service or self-managed, version, instance class, storage and HA design
Schema and data-type conversion
Including the NUMBER, DATE and VARCHAR2 decisions that change application behaviour
PL/SQL to PL/pgSQL conversion
Packages, cursors, exceptions, autonomous transactions
Application SQL refactoring
Oracle-only syntax, drivers, connection strings and ORM dialect
Data migration with change data capture
Full load followed by CDC, so the cutover window is short and planned
Row-level reconciliation and functional testing
Row counts, checksums and query-result comparison against Oracle
Performance tuning on PostgreSQL
Plan review, index strategy, statistics, connection pooling
Cutover and rollback plan
Rehearsed, with a documented path back to Oracle
Post-cutover support
Hypercare through the first weeks in production
How the migration works
The same four phases we run on every database migration, with what each one means on an Oracle source.
Discover
Inventory tables, views, materialized views, packages, procedures, functions, triggers, sequences, types, synonyms, DB links and scheduler jobs. Grade the PL/SQL by conversion complexity rather than by line count. Identify the Oracle-only features in use, because those decide whether the migration is routine or not.
Design
Choose the target, the PostgreSQL major version and the sizing, then the cutover pattern: a one-time cutover with writes paused, or a phased migration with change data capture holding both systems in step.
Transform
Kanetic-x translates around 90% of PL/SQL to PL/pgSQL automatically, alongside the schema and data types, and a database engineer reviews every converted object before it reaches a target schema. Packages with state, autonomous transactions and optimizer hints sit in the remaining share by definition rather than by accident, and are redesigned by hand.
Validate
Row counts and column checksums per table, query-result comparison for the reports the business actually reads, and a performance baseline against Oracle. The cutover and the rollback are both rehearsed before the real one.
Want the engineering detail? Read our step-by-step Oracle to PostgreSQL migration guide, which covers each phase with the SQL.
Where the effort really sits in an Oracle to PostgreSQL migration
Tables and data convert almost for free. These ten Oracle features are where the days go, and roughly half of them cannot be automated at all.
| Oracle feature | PostgreSQL approach | Typical handling |
|---|---|---|
| Packages (with package state) | Functions grouped in a schema; state via session settings or temporary tables | Manual redesign |
| Autonomous transactions | dblink or pg_background extension | Manual redesign |
| Global temporary tables | Temporary tables created per session, or unlogged tables | Partly automated |
| CONNECT BY hierarchical queries | WITH RECURSIVE | Mostly automated |
| (+) outer-join syntax | ANSI JOIN | Automated |
| ROWNUM | LIMIT, or ROW_NUMBER() where the ordering matters | Automated |
| Synonyms | Views or search_path | Partly automated |
| DBMS_* and UTL_* packages | orafce extension, other extensions, or logic moved into the application | Case by case |
| Optimizer hints | Not supported natively; pg_hint_plan or a query rewrite | Manual tuning |
| Empty string treated as NULL | PostgreSQL keeps '' distinct from NULL, so the logic has to be read | Manual review |
Kanetic-x translates around 90% of PL/SQL automatically. The rows marked manual are where the other share concentrates, and they are redesigned rather than converted. "Typical handling" describes what tooling generally covers, not a guarantee for a specific codebase.
The guide works through how each of these is converted, with SQL examples.
Oracle to PostgreSQL on AWS, Azure and Google Cloud
Each platform has its own native migration tooling, and each has a gap the tooling does not cover. Knowing the gap in advance is most of the planning.
Oracle to PostgreSQL on AWS
We land you on Amazon RDS for PostgreSQL or Amazon Aurora PostgreSQL, using AWS Schema Conversion Tool for the schema and code and AWS Database Migration Service for the full load and the ongoing CDC. The gap worth knowing before you plan: AWS states that DMS "doesn't migrate your secondary indexes, sequences, default values, stored procedures, triggers, synonyms, views, and other schema objects not specifically related to data migration" and that AWS SCT is what handles those. AWS also documents a rollback for this path, which works by reversing replication back to Oracle, so it only exists if the reverse task was built during cutover. AWS migration walkthrough.
Azure Oracle to PostgreSQL migration
The target is Azure Database for PostgreSQL flexible server. Microsoft's current documented path converts the schema and the Oracle-dependent application code in the PostgreSQL extension for Visual Studio Code, which uses Microsoft Foundry models, and then copies the row data. Note that Azure Database Migration Service does not list Oracle as a source for the PostgreSQL target, so this is not a single-service migration. Where an open-source path suits the estate better, Ora2Pg converts the schema and code and moves the data, with no CDC. Microsoft's Oracle data migration overview.
Oracle to Postgres migration on GCP
Targets are Cloud SQL for PostgreSQL and AlloyDB for PostgreSQL, both supported on versions 12 through 18. Google Cloud Database Migration Service handles heterogeneous Oracle migrations with conversion workspaces, and for AlloyDB those workspaces add Gemini-assisted code explainability and conversion-issue fixing. Supported Oracle sources are 11.2.0.4, 12.1.0.2, 12.2.0.1, 18c, 19c and 21c. Oracle Autonomous Database is not supported as a source, which rules the service out for some estates entirely. Google's supported sources and versions.
Self-managed or on-premise PostgreSQL
Chosen for data residency, for an existing on-premise estate that is not moving, or because the managed-service pricing does not work at your scale. The work that a managed service would have done for you becomes part of the project: streaming replication and failover, backup and point-in-time recovery, patching, and connection pooling with PgBouncer, which matters more on PostgreSQL than on Oracle because PostgreSQL runs a process per connection.
Where the database move is part of a wider platform change, it runs alongside our cloud migration services.
Not sure how much of your PL/SQL will convert automatically?
The assessment inventories the Oracle objects, grades the PL/SQL by conversion complexity and names the Oracle-only features in use, so the estimate rests on your code rather than on an average. It takes one to two weeks depending on the size of the estate.
Why teams choose Kansoft for Oracle to PostgreSQL migration
AI does the first pass, engineers own the result
Kanetic-x translates around 90% of PL/SQL on a first pass and generates tests against the original behaviour, and a database engineer reviews every converted object before it reaches a target schema. The tooling changes how long translation takes. It does not change who is accountable for the result.
A named US accountability point
Engineering runs from our delivery centres in India, with a director based in Lexington, Kentucky who owns the relationship in US hours. You are told which model you are buying before you sign, not after the first escalation.
We will tell you not to migrate
The assessment is scoped and priced separately from delivery, which is what makes it possible for it to conclude that a database should stay on Oracle. Four of those situations are listed further up this page. Delivery runs against ISO 27001 and SOC 2 Type II aligned controls.
Frequently asked questions about Oracle to PostgreSQL migration
The questions that decide whether a migration from Oracle to PostgreSQL gets approved.
How much does an Oracle to PostgreSQL migration cost?
How long does an Oracle to PostgreSQL migration take?
Can we migrate without downtime?
Will our applications need to change?
Is PostgreSQL reliable enough for enterprise workloads?
Can we migrate one database or application at a time?
Do you only use Kanetic-x, or also AWS SCT, Ora2Pg and Google DMS?
What do we get from the free assessment?
Get your free Oracle to PostgreSQL migration assessment
Tell us the Oracle version, roughly how large the database is and where you want to land. You will talk to a database architect, not an account manager, and the output is an object inventory with a graded conversion estimate rather than a proposal. It takes one to two weeks depending on the size of the estate.
Related
Last updated 2026-11-18