Oracle's in-memory database: Ripe for the picking by open source

Though snazzy and 'ungodly' fast, Oracle's in-memory database will inevitably give way to open source implementations

Last night, Larry Ellison kicked off Oracle Openworld 2013 with a keynote speech that highlighted one of Oracle's highly touted new technologies: in-memory database processing for "ungodly" speed (Ellison's word).

For all its wow and dazzle, though, this technology, which has already been deployed by other big-league database vendors, is only likely to remain a major draw for Oracle until an open source implementation comes along.

The big advantage of processing databases entirely in-memory is speed, as the Oracle demo showed off. Databases for common line-of-business applications run to many dozens of gigabytes or more, though, which means for a long time, such processing techniques have been confined to the most elite and expensive of database systems.

But as Ellison pointed out, memory is cheap and getting cheaper, which makes the in-memory approach far more practical than it used to be. It also makes that approach something far easier for open source competition to eventually produce.

Two major considerations to keep in mind in light of the Oracle demo:

1. In-memory database technology is far from new

The fact that Ellison can invoke in-memory databases as a buzzword is hint enough that the technology itself isn't something Oracle invented, let alone a new technique at all.

The fact is, Oracle's major competition has been providing similar solutions for some time now. Sybase, for instance, has been offering an in-memory database solution of its own, ASE, since 2010; SAP has HANA. (Both are implemented differently and to different ends -- ASE is for OLTP; HANA is more for BI plus OLTP -- but the core is the same.) IBM's solidDB product, acquired from a third party, has been providing in-memory database tech since 2008.

The main ingredient in Oracle's secret sauce for this offering is how it's bringing in-memory databases to its own customers as a finished product, not just as a raw technology. But that can't last forever, either.

2. A full-blown open source implementation can't be far off

The history of software shows most every proprietary application is soon followed up by an open source equivalent. In the database world, MySQL/MariaDB and PostgreSQL have shown themselves to be worthy contenders to proprietary offerings from Oracle, IBM, Microsoft, and many others.

What's more, the technology to provide in-memory databases is available in the open source world. A product like VoltDB is one example, but the major names have the pieces for such behavior under their hoods right now. MariaDB (and its Oracle-owned predecessor, MySQL) has an in-memory database engine; likewise, PostgreSQL has similar functionality.

Right now, none of these are drop-in solutions. If a database maven wanted to turn them into a full-blown in-memory data solution, he'd have to do a great deal of heavy lifting.

That's where Oracle still has an edge: By being able to deliver a polished product that works as a one-to-one upgrade for existing solutions, albeit at a prime price.

But Oracle doesn't even have the monopoly on such things now, and a full in-memory edition of one of the existing enterprise-level open source database would give people that many less reasons to stay with Oracle.

This story, "Oracle's in-memory database: Ripe for the picking by open source," was originally published at Get the first word on what the important tech news really means with the InfoWorld Tech Watch blog. For the latest developments in business technology news, follow on Twitter.

From CIO: 8 Free Online Courses to Grow Your Tech Skills