Adaptive Quotient Filter (AQF)

The AQF supports approximate membership testing and counting the occurrences of items in a data set. Like other AMQs, the AQF has a chance for false positives during queries. However, the AQF has the ability to adapt to false positives after they have occurred so they are not repeated. At the same time, the AQF maintains the benefits of a quotient filter, as it is small and fast, has good locality of reference, scales out of RAM to SSD, and supports deletions, counting, resizing, merging, and highly concurrent access.

• 'qf_insert_ret(item, count)': insert an item to the filter. Returns 0 if the item to insert shares a fingerprint with an existing item in the filter.
• 'insert_and_extend(item, count, other)': insert an item and extend it. Should be called after qf_insert_ret returns 0, where other is the key of the previously inserted item which caused qf_insert_ret to return 0. If item and other are equal, this updates the count for the item in the filter.
• 'qf_query(item)': return the count of the item. Note that this method may return false positive results like Bloom filters or an over count.
• 'qf_remove(item)': remove the item from the filter
• 'qf_adapt(item, other)': extend the item in the filter. Afterwards, other is guaranteed to not be a false positive query unless the two items are identical.

This library depends on libssl.

The code uses two new instructions to implement select on machine words introduced in intel's Haswell line of CPUs. However, there is also an alternate implementation of select on machine words to work on CPUs older than Haswell.

To build on a Haswell or newer hardware:

 $ make test
 $ ./test

To build on an older hardare (older than Haswell):

 $ make NH=1 test
 $ ./test

Contributions via GitHub pull requests are welcome.

• Prashant Pandey [email protected]
• Rob Johnson [email protected]