[supergirl]

Don't know what you consider simple, but databases use bloom filters to do faster joins for example. Chrome also uses them for malware detection I think.

[throwaway1979]

Specifically, a place where DBs use bloom filters is distributed joins. This isn't a common scenario IMHO. Found a good description here:

http://www.coolsnap.net/kevin/?p=19

P.S. Loved the tutorial.

[supergirl]

it's also useful on local hash joins because it's faster than a regular hash table lookup. so you can use it to eliminate some stuff before going to the big hash table.

[llimllib]

> Chrome also uses them for malware detection I think.

Used to! I had them in earlier versions of the article but they changed that code.

[im3w1l]

Do you know what they changed them to and why?

[llimllib]

Used to! But it was quite a while ago that I noticed that they had changed it, I don't remember any longer.

[fennecfoxen]

To generalize, "possible match detected" for anything: you have a big, fat database about a bigger, fatter universe of identifiers, and a less-capable set of clients over a less-capable network link. They want a fast way of determining whether or not a particular identifier is worth querying the database for, and the false positives do not have a very high cost.

The conditions are relevant for some cloud-managed appliances, mobile-phone applications, and sometimes caching techniques in your own distributed applications.