Why Oracle Database Never Had a Version 1

Pull your chair a little closer. The fire is warm tonight.

Here is a question I guarantee has crossed your mind at some point. Maybe while reading the docs. Maybe during a version upgrade. Maybe in the middle of a conversation you barely remember.

Why does Oracle Database start at Version 2?

There is no Oracle Database Version 1 in the official release history. The product launched commercially at Version 2 in 1979, as if the number one had simply been skipped. No explanation in the manual. No footnote. Just Version 2, first release, end of story.

Except it is not the end of the story. It is the beginning of one.

The Paper That Sparked a Revolution

In 1970, a British computer scientist named Edgar F. Codd published a paper called A Relational Model of Data for Large Shared Data Banks. Dense. Mathematical. The kind of thing that most people read past without stopping.

Three people stopped.

Larry Ellison, Bob Miner, and Ed Oates were running a small consultancy in 1977. They read Codd’s paper and spotted what the established players had largely overlooked: a blueprint for something that did not yet exist commercially. A true relational database, queryable through a structured language, ready for enterprise use.

The biggest names in the industry had been sitting on the concept for years. Moving carefully. Filing patents. Running proofs of concept in research labs.

Ellison and Miner decided to actually build it.

They named it Oracle, borrowed from the codename of a classified government project Ellison had previously worked on. They wrote the first version in assembly language and ran it on a minicomputer with just 128 kilobytes of memory. If you have ever tried to open a single browser tab on 128KB, you understand what kind of engineering that required.

They finished it in 1978. They called it Oracle Version 1.

And then they made a decision that would echo for the next forty-plus years.

What Was Oracle Database Version 1, Actually

Oracle Database Version 1 existed. It ran. It processed queries on hardware that most developers today would not recognise. It was a working proof of concept that showed Codd’s relational model could function in practice, not just in theory.

But it was rough.

The team knew it. The code was functional but not something you would hand to an enterprise customer with a straight face. It was a prototype, the kind of thing you build to convince yourself the idea works before you convince anyone else.

So they kept building. Refining. Cleaning up the implementation. By 1979, they had something they could actually stand behind and sell.

And here is where it gets interesting.

The Day They Skipped a Number

In 1979, Ellison made the call.

They would launch as Oracle Version 2.

Not as the first public release of a promising prototype. Not as version 1.0, some features pending. As Version 2. The reasoning, according to Ellison himself, was direct: he knew no one would want to buy Version 1.

Think about that from a customer perspective for a moment.

It is 1979. Enterprise software is expensive and the people signing purchase orders are not developers. They are managers betting their careers on these decisions. A small consultancy you have never heard of walks in trying to sell you database infrastructure. The product is called Version 1.

Version 1 means you are the experiment. You will find the bugs. You are the test case.

Version 2, though? Version 2 has a history. Version 2 implies someone already survived Version 1. The edges are smoother. The risk is lower.

It was, objectively, a fiction. There were no Version 1 customers. Oracle Database Version 1 had never left the building. But the signal that Version 2 sent to the market was worth more than the literal truth of the version number.

Why That Number Mattered

I have been building Oracle APEX applications long enough to know that how you present something shapes how it gets received. Not dishonestly. Honestly, but strategically.

The Oracle Version 2 decision is the earliest and most famous example of this in our world. Ellison did not misrepresent what the software did. According to Oracle’s official documentation on the history of Oracle Database, Version 2 was the first commercially available SQL-based relational database management system in the world. The version number just told a more confident story about where it had come from.

That instinct matters whether you are shipping enterprise software, publishing a blog, building an APEX plugin, or presenting your work to a client. The craft is not in the deception. It is in understanding that perception shapes adoption as much as the product itself does.

If you are figuring out how to start your Oracle developer career in 2026, that lesson is worth sitting with early.

The Legacy That Followed

Oracle Version 2 sold. Then Version 3 arrived in 1983, rewritten in C, making it the first relational database to run on mainframes, minicomputers, and personal computers on a single codebase. Version 4 introduced multiversion read consistency. Version 5 brought client-server architecture. Version 6 gave us PL/SQL.

The PL/SQL we still write today. Every block, every procedure, every function you have debugged at 11pm. That lineage starts right here.

If you want to understand the full shape of what Oracle Database has become, from the different types of Oracle Database available today to the cloud migration path modern teams are taking, that history is the foundation underneath all of it. And if you want to understand why Oracle APEX holds the position it does in the low-code world, the answer is the same: decades of exactly this kind of pragmatic, fast-moving engineering.

The knowledge of these stories is what separates developers who understand Oracle from developers who just use it. The Oracle ACE community runs on this kind of institutional memory. Find the people who carry it. Ask them questions. They will tell you things no documentation covers.

As for Oracle Database Version 1, it lives somewhere in the archive. Written in assembly language, running on hardware that belongs in a museum, processing SQL that barely resembles what we write today. It proved the idea was possible.

Version 2 proved it could sell.

---

The fire is still going. Same time next month, last Friday of July.