DuckDB, database wrangler used by Google, Facebook, and Airbnb, hits 0.5.0

DuckDB – the in-process analytical database management system used by Google, Facebook, and Airbnb – has released its 0.5.0 iteration. The brainchild of academics at Amsterdam's Centrum Wiskunde & Informatica mathematical and theoretical computing research center, DuckDB is embedded within a host process. There is no DBMS …

[Tom 7]

Wilbur

is not a Call Duck. Peak Guardian again!

[Anonymous Coward]

Re: Wilbur

It most certainly is a Call Duck, as recognised by the British Waterfowl Association.

[Tom 7]

Re: Wilbur

Call Ducks are white occasionally with a bit of colour - the one in the Guardian picture is a Mallard. Call ducks have a different morphology.

[Anonymous Coward]

Re: Wilbur

Peak dumbass Guardian critic.

"The colours represented in Call Ducks are derived from the same colour genotypes as their larger relatives. "

https://www.waterfowl.org.uk/domestic-waterfowl/call-ducks/

[Cederic]

Re: Wilbur

This is why I love this site. Expertise and insight on all technology matters, then well beyond.

The key question not answered: Do Call Ducks quack? I went somewhere in the US and they were proud of their ducks that don't quack and.. who wants ducks that don't quack!

I'd like a pet duck, I'm just not sure my cats would.

[Korev]

I've been using it for a while now and rather like it. It generally performs better than SQLite and being able to query straight from Parquet is useful.

One major PITA is at least in the previous Python bindings indices are ignored if you use a parametrised query meaning you pretty much have a choice between performance and security[0]

I'm also not a fan of the breaking changes with each version <1 having breaking changes.

A pint to the team to celebrate their new release -->

[0] You can of course write your own function to sanitise your inputs, but you shouldn't have to.

[RichardB]

I'm imagining it a something like a local and stripped version of Apache Drill.... Am I close?

If so sounds pretty handy.

[Tom 7]

I've had a wander around the site but finding it difficult to find decent use cases. My worry is in the wrong hands its will lead to massive tables and the grinding of gears as indexes are updated. There are reasons why OLAP has gone a different way before now.

[Anonymous Coward]

I'm only using a hobbyist ~40 million meta record set for audio files/songs but, I found no differences in read only queries between DuckDB and SQLite. I only really care about SQL read speed so I'm under the impression that column based storage is faster but I've yet to find anything faster than SQLite :-/

It would be really nice in general if "FOSS" DB softwares would demonstrate the min/max record complexity and the same for tables/rows, at least for read operations. I don't mean O notations as those vary across CPU's, I mean actual real world data/record sets given for examples. Sure it can be said "it depends" and of course the record set could be rigged to a bias but, the sooner I have a working record set the sooner I can judge for performance and biases.... give me something to test run!!!

[RichardB]

I'd guess it allows you to tap into Jason's, parquets, Csvs etc with just a straight SQL query

[spireite]

If it proves better than all the others .

.. then the competition's goose is cooked?