Hasso Plattner, SAP’s co-founder and current supervisory board chairman, has been known to fly high in his keynote address at the company’s annual Sapphire user conference. This year, Plattner took a high-flying concept, the growing in-memory database market, and brought it down to earth for all to see. And what attendees saw was the beginning of a major revolution in database design, cost, and performance — and the beginning of a major headache for anyone addicted to the aging relational database model.

At the core of Plattner’s mini-lecture on databases were two not-so-new technologies that SAP has been busy blending together. The first is the column-based database, a method of storing and accessing data that has two major benefits: It requires much less space to store a given amount of raw data than a relational database, and it is self-indexed, eliminating the overhead for indexing, which can more than double the size of a relational database. Column-based databases are also dead-simple to update, both in terms of their structure and their content, unlike a relational database, which needs to be “reprogrammed” every time its structure changes.

SAP has been moving this column-based capability into memory, the second not-so-new technology playing a central role in SAP’s latest effort. Plattner showed off a $6,000 rack-based system with 144 gigabytes of RAM. That’s enough RAM to load up a column-based database that would require 2 terabytes as a relational database.

And then the fun began. Plattner demonstrated SAP’s new Explorer analytical interface in order to show how fast this in-memory, column-based database can crank out analysis. Explorer has some lovely, natural-language features, but its most impressive feature was its ability to process 280 million records in under a second, powered by SAP’s column-based, in-memory database. For anyone trying to analyze massive quantities of data, it was an impressive show. (For the record, it can process 400 million records in under a second, though that was not what the demonstration showed.)

Then Plattner unleashed the real revolutionary idea. While the advantages of column-based, in-memory databases have been known for some time, this one-two punch has always been considered appropriate only for the analytical side of the house, as a replacement for a data warehouse. No one has publicly suggested, at least no one at SAP, that this same technology could be used to power the core transactional system at the heart of SAP’s Business Suite.

And that’s just what Plattner did. “Why not use this for transactional systems?” he asked, revealing SAP’s intentions.

Though there was no timeline offered for when this technology would be available to run the core of SAP, it’s clear that the vendor could start prototyping this today at a customer site. At which point every relational database vendor — and every relational database administrator — had better start worrying. SAP’s ERP customers are some of the higher-profile consumers of relational technology, and while there has been little incentive for them to ditch Oracle, IBM, or Microsoft to date, this new in-memory, column-based approach presents a compelling argument for change.

Revolutions don’t happen overnight. But the seeds of the new database revolution, planted years ago, are starting to sprout. SAP may have found the weapon it needs to unseat Oracle as the number one database in the SAP customer base, and in the process make Oracle’s database look a little slow and tired — at which point we can anticipate that Oracle will kick off its own database revolution.