5 comments

[ 3.5 ms ] story [ 18.7 ms ] thread
Hmm, just looked through the slides (listed to the video at the same time).

The only interesting slide is the last slide VoltDB (new kid on the block), promises some nice performance..., but that's the only slide that actually talks about it (out of 30?) and does not really support it. So HN adpost ?

Isn't Oracle able to do in-memory and columnar now?
Current VoltDB, former Vertica dev here. The simplest answer to this question is basically to argue that the architecture and implementation matters, and that Oracle (and others) are often bogged down by compatibility constraints and conservatism to really compete on a performance level with in-memory systems like VoltDB, or columnar systems like Vertica.

It's dangerous to underestimate them, because they have money, smart engineers and a machine of a salesforce, with support and relationships that help them sell to lots of customers. But they don't keep me up at night, and most of the deals we see Oracle in (at VoltDB) start with the sentence, "We're looking at TimesTen replacements."

A key thing here is Oracle also costs a truckload of money. Price pressure on Oracle as startups and Open source (or combos) mature is an interesting thing to watch.