Here's an FAQ about HP-Oracle's Database Machine and Exadata Storage Server:

Which one is the "database accelerator" that Ellison had been teasing us about?

The Exadata Storage Server, a standard rack-mountable HP ProLiant DL180 G5 server sporting two Intel quad-core CPUs connected to 12 hard drives of 1TB each.

What makes it different than a typical Linux-based storage server is the fast parallel-query software built into it, which allows the Exadata to perform a number of database functions locally. The Exadata can "optimize queries" by doing lower-level calculations closer to the raw data. Only the results are then sent to the actual database, which aggregates the results in order to perform the final number crunching.

"The intelligence allows us to reduce the amount of data flowing across the interconnect," said Ellison, which, as he rightfully pointed out, is the source of most of the delay in modern database systems.

Besides reducing the data being transmitted through the network, the Servers are also equipped with two InfiniBand connections for high-speed data transfer. Ellison says Exadata users can expect a real-world bandwidth today of 1Gbit/sec, which he claimed is far faster than conventional disk storage arrays.

And the Database Machine?

The imposingly named Database Machine is essentially an all-in-one database and storage solution for companies seeking building blocks as their databases near the petabyte range.

The Database Machine comes on a single rack and includes 14 Exadata Storage Servers, eight Linux servers running Oracle Database 11g Enterprise Edition, a total of 368GB of RAM and 168TB of disk space.

Ellison joked, "This holds a lot of songs. It's 1,400 times larger than Apple's largest iPod."

Not only is it massive; Oracle also claims it's ultrafast because the machine brings the database servers even closer to the Exadata servers in a network with a raw I/O of 14Gbit/sec.

How fast is the new breed?

Ellison said the Database Machine was between 10 and 72 times faster than conventional Oracle databases. A number of Oracle customers have been testing the Machine for a year, putting their actual production workloads onto half-sized Oracle Database Machines ("because we're really cheap," quipped Ellison).