There's also the In Memory Columnar Store [0, 1] by Konstantin Knizhnik, if anyone's interested.

[0]: http://www.pgcon.org/2014/schedule/events/643.en.html

[1]: http://www.garret.ru/imcs/user_guide.html

Compared to Monetdb:

https://www.monetdb.org/content/citusdb-postgresql-column-st...

I briefly worked for a place that had been through cstore and greenplum before buying into Vertica, right before I started. They were simply hungry for analytical performance and had the cash. It was a really great buy for them in performance but also ease of use.

Hindsight: I was skeptical that they'd left all the metadata (configuration/application domain data) in PostgreSQL while moving the (100x larger) business data to Vertica. At this point I'd repeat the pattern just for the sheer delight of dealing with a complex application schema in PostgreSQL. The flexibility demanded of Vertica was decidedly less in that shop.

Did you guys perhaps benchmark kdb vs. Vertica?

Nope. Vertica provided the right performance and the right (SQL) interface for analysts.

This isn't the same cstore as in Stonebraker et al, right? http://db.csail.mit.edu/projects/cstore/

Nope, they're two different systems with a similar name.

This deserves a closer look. It appears to be making use of the ORC format used in Hive, and last I looked was implemented in Java and tied in tight with the Hive arch. So this must be a rewrite of an ORC reader/writer. Selfishly I wish they'd picked Parquet instead of ORC.

This looks very promising

Excited! Let's see how it performs.