Some column stores are not like other column stores

[dvryaboy]

This is a great collection of pointers!

The current list of column stores conflates two things: "key-map" data stores like Cassandra, BigTable, and HBase and the rest -- the majority of the list -- and actual column stores like C-Store, MonetDB, and the Parquet file format.

These two designs are aimed at very different workloads and are fundamentally different structurally; they should be broken out into different categories.

Unfortunately both groups have traditionally called their technology "columnar databases", while meaning completely different things, so there isn't a great term to use for one vs the other.

I've seen "wide-column key-value datastores" used to describe BigTable and friends; this is definitely more technically correct, though I am sure it's confusing to the reader who now has to find out what the difference is between that and a column store. But perhaps that's a good thing.

Prof. Dan Abadi has a few other naming suggestions, as well as a detailed explanation of why these things are fundamentally different, here: http://dbmsmusings.blogspot.com/2010/03/distinguishing-two-major-types-of_29.html

[oxnr]

Sounds great. Feel free to create a pull request with the separation. Thats exactly why i put this list on github to put more brains into this topic ;)