Oracle to PostgreSQL TCO: a realistic cost-savings model for 2026

If you have ever sat in a budget meeting where the question is "what does moving off Oracle actually save us," you know the back-of-the-envelope numbers vary by an order of magnitude. The vendor-friendly version says you save 60 to 80 percent on database costs. The realistic version says it depends on how much PL/SQL you have, how disciplined your migration is, and which target you pick.

This post lays out a working TCO model. Plug in your numbers, get a real estimate, and bring it to the next budget meeting.

The five cost categories

A complete TCO model has five line items:

1. Database license + support (Oracle vs Postgres)
2. Hardware and infrastructure (per-core licensing affects this on Oracle)
3. DBA and operations time (Oracle DBAs are scarce and expensive)
4. Migration cost (one-time)
5. Application changes (one-time, usually smaller than expected)

Let us go through each.

1. Database license and support

This is the line item everyone shows in the slide deck. It is also the line item most likely to be misquoted.

Oracle

Oracle Database Enterprise Edition lists at $47,500 per processor (Named User Plus pricing is different). Standard Edition 2 lists at $17,500 per processor. Annual support is 22 percent of license, so a fully licensed-and-supported Enterprise Edition core costs roughly $58,000 in year one and $10,450 each year after.

Real customers usually negotiate down significantly on the license but never on support. A 16-core deployment is a $928,000 list-price commitment, with $167,200 per year in support after.

PostgreSQL

PostgreSQL is open source. The license is free. The cost is in the support contract you choose:

• Self-supported: $0
• EnterpriseDB or Crunchy Data 24x7 production support: roughly $5,000 to $20,000 per cluster per year, depending on tier
• Professional services for migration cutover: separate

For a 16-core deployment supported by a commercial vendor at the higher tier, you are looking at $20,000 to $40,000 per year. Compared to Oracle's $167,200 in year-two-onward support alone, the savings are dramatic and well-documented.

Year-one license + support savings on a 16-core deployment: roughly $890,000.

2. Hardware and infrastructure

Oracle's per-core licensing makes hardware decisions political. You buy fewer cores than you actually need because each core costs $58,000. PostgreSQL has no per-core fee, so you size for actual workload.

In practice, customers moving off Oracle usually:

• Add 30 to 50 percent more compute on the Postgres target (because they finally can without a tax)
• Move from on-prem to cloud (AWS RDS for PostgreSQL, GCP Cloud SQL, Azure Database for PostgreSQL, or self-managed on EC2/GCE)
• Eliminate Oracle-specific hardware contracts (Exadata, ZFS Storage)

The net infrastructure spend often increases slightly while throughput doubles. The economic comparison is per-transaction, not per-core.

3. DBA and operations time

Oracle DBAs are expensive and scarce. Median total compensation in the US ranges from $145,000 to $185,000 depending on seniority. PostgreSQL DBAs trend lower: $115,000 to $155,000.

More importantly, PostgreSQL DBAs are roughly 4x more available in the labor market. If you are running a team of 6 Oracle DBAs and you cut to 4 PostgreSQL DBAs (which is realistic for the same workload), you save $300,000 to $500,000 per year in labor.

We are not advocating cutting headcount. Most migrating teams redirect those people to higher-leverage work like data platform engineering. But the math is real, and finance will ask.

4. Migration cost (one-time)

This is the line item where every estimate is wrong, and almost always too low.

The traditional path

A typical Oracle-to-Postgres migration with hand-written conversion:

• 6 to 18 months of elapsed time
• 2 to 6 senior database engineers full-time
• Specialized professional services contract: $300,000 to $1,500,000
• Real-world delivery rate: 60 to 70 percent automated, 30 to 40 percent manual SQL editing on the long tail

A 30,000-object estate (typical for an enterprise ERP, banking, or insurance system) at 65 percent automated conversion leaves 10,500 objects for engineers to translate by hand. At 30 minutes per object on average, that is 5,250 engineer-hours, or roughly $750,000 in fully-loaded labor.

Total typical migration cost: $1.0M to $2.5M, 12 to 24 months elapsed.

The Trekport path

Trekport's design goal is 100 percent automated conversion with zero manual SQL editing. The cost model becomes:

• Trekport Studio: $49 per month per developer for the trial scope, $299 per seat per month for enterprise teams with unlimited data transfer
• 1 to 3 months of elapsed migration time on a typical 30,000-object estate
• 1 to 2 engineers full-time during the migration window
• No external professional services for the conversion itself (you may still want them for cutover testing)

Total Trekport-path migration cost: $50,000 to $300,000, 1 to 3 months elapsed.

The differential is not a 2x or 3x improvement. It is an order of magnitude. And the timeline savings often matter more than the dollar savings: every month you delay cutover is another month of paying both Oracle support and the migration team.

5. Application changes

The application code that talks to your database needs review. The good news: most of it is portable.

• Standard SQL queries: usually compile against PostgreSQL unchanged. Estimated effort: 5 percent of files need touch-ups.
• Oracle-specific functions in queries (NVL, DECODE, SYSDATE, INSTR): if you install orafce on the Postgres target, these run unmodified.
• Cursor-based ORM mappings: most ORMs (Hibernate, SQLAlchemy, Active Record) have first-class PostgreSQL support and need only connection string changes.
• Stored procedure calls from application code: signatures are preserved by Trekport's package-as-schema convention, so call sites usually work unchanged.

Realistic application-side effort for a typical enterprise application: 10 to 30 engineer-days of touch-ups. Call it $20,000 to $80,000.

Putting it all together

For a representative mid-market enterprise running a 16-core Oracle Enterprise Edition deployment with a 30,000-object estate:

| Category | Year 1 (Oracle) | Year 1 (Postgres after migration) | Annual savings (steady state) | |---|---|---|---| | License + support | $928k initial + $167k/yr | $20k to $40k/yr | ~$140k/yr | | Hardware | Status quo | +20% capacity, similar cost | $0 | | DBA labor | $1.0M (6 DBAs) | $560k (4 DBAs) | $440k/yr | | Migration cost (Trekport path) | n/a | $50k to $300k one-time | n/a | | Application changes | n/a | $20k to $80k one-time | n/a | | Net | ~$2.1M/yr | ~$1.6M one-time + $620k/yr | ~$580k/yr ongoing |

Year one breakeven is roughly month 7. Year two through five savings are roughly $580,000 per year. Five-year savings against staying on Oracle: roughly $2.5 million.

Your numbers will differ. The point is that the math is positive even on conservative estimates, and the migration tooling line item ($50k to $300k with Trekport, vs $750k+ with hand conversion) is where most of the variance lives.

What people miss

Two things customers consistently underestimate when modeling this:

1. The cost of partial conversion. If you use a tool that gets you 70 percent automated, you still owe the 30 percent in engineering time. That 30 percent is not 30 percent of the easy work. It is 100 percent of the hard work: packages, compound triggers, dependency-ordered objects. Plan for it explicitly or do not.
2. The cost of running Oracle and Postgres in parallel during cutover. Most migrations run both databases for 3 to 9 months while the application team validates against the new target. That is double licensing during the overlap. Faster migration tools shorten this overlap and save you the duplicate cost directly.

Try the calculator

Want to plug in your real numbers? Talk to us and we will share the spreadsheet we use with prospective customers. It is more granular than the model above, breaks out cost by year, and includes sensitivity analysis for the migration tooling line item.

If you are ready to move, download Trekport Studio and run the free trial against a representative subset of your Oracle schema. The conversion report tells you, before you commit to anything, what 100 percent automated conversion looks like for your specific estate.