Ovrimos logo

From 1996 to 2000

This is a story few people already know, but some people convinced me it was time for it to be told to a wider audience.

Ovrimos was a Web-enabled SQL Server, the only one ever built in Greece, as far as I know. It all begun in 1996...


Ovrimos was built by a company named Altera (no connection with the same-named American company), which I joined in 1995. The company was making a living by creating custom DOS applications with a proprietary DBMS, based on a BTree library built by one of the owners. At the time, the competition was mainly Oracle Forms applications which were crawling like a dying salamander in a tar pit. Altera's applications, on the other hand, using explicit iterations on BTrees and Spartan DOS UIs, where blazingly fast and appreciated by its customers.

However, in a stroke of genius, or madness, or both, the company decided to move into product development, and develop the first Greek RDBMS. The owner used to recite a (fictitious?) story about a big software company and how they developed a revolutinary O.S. using junior newbies fresh out of school who succeeded because they had no idea that what they were doing was considered extremely difficult. That was what he was planning to do.

The product was called "Altera SQL Server", at first, but then the American Altera must have put some pressure on the company, because it was renamed to "Ovrimos" (meaning "brave" in ancient Greek) and "Ovrimos S.A." was created in the U.S. to represent it. There were essentially two major releases (2.0 and 2.5) until my departure in December 2000. I believe the difference between the two was the presence of the DatabaseManager, which was a program managing database instances, and the incorporation of OpenSSL to enable HTTPS. I think the Replicator (which transmitted transactions to a remote, read-only, Ovrimos server) was incorporated into version 3.0, which was released after my departure and also sported stored procedures and functions in SQL (version 2.5 supported stored procedures through a CGI interface, or written in Scheme).


At the time we started development, apart from myself there was only one senior programmer in the team, Giorgos Georgopoulos, and one or two complete juniors. Later on, Giorgos left the company. Miltiadis Kokkonidis joined Altera and worked on the Database Manager, the OpenSSL integration and a custom Servlet implementation to embed into Ovrimos (which was finally not incorporated) before he left to follow post-graduate studies in the U.K.. At some point my friend, Jiannis Tzikas, joined Altera to work on the Replicator, the ODBC 2.0 driver and the TPC-C testing.

Package contents

Ovrimos was a group effort, and I hope the few other people involved will also write an account of it from their own viewpoint. But, as the person having written (or inherited and maintained) most of the code up until release 2.5, and all of the core components, I consider Ovrimos my spiritual child. It was a great school in the technical realm. Of course, it meant I spent five years of my life living in a cave, so to speak, disconnected from the realities of usual software development in a corporate setting. But it's an experience that marked me, and I wanted to share it with you.

What went wrong

Ovrimos was not a toy system. Much effort went into its becoming stable and scalable. We ran TPC-C tests with custom stress-test tools that also checked validity of the results. We had nested transactions when MySql was introducing experimental support for per-table transactions. We had a Web UI when the Web was, oh, so new. What went wrong with it?

Although we ported to almost all widely available Unixes of the time, and all business opportunities that were pursued involved Linux or BSD, the management seemed to be chasing after the market segment of Oracle and SQLServer. They did not succeed. The company went belly-up.

They tried to bundle it with various BSD and Linux distributions (and also the Cobalt internet appliance), with the goal that users would upgrade from the free 1-user license (same one in the demo) to a paid, less restrictive one. No revenue came from that.

It is my belief that Ovrimos would have made a great replacement for MySql and a great database for Linux and assorted Unixes. I had proposed the venue of open-sourcing it, but there was no clear business plan at the time of how to make money by giving away open source software.

Proceed to read about Technology Proceed to Technology