PFCOUNT

PFCOUNT is a Redis command that returns the approximate cardinality of the union of HyperLogLog structures stored at one or more keys. A HyperLogLog is a probabilistic data structure used to estimate the number of distinct elements in a set while using a fixed, small amount of memory. PFCOUNT takes one or more keys as arguments and computes an estimate of the total number of distinct elements across the provided HyperLogLogs. If a key does not exist, it is treated as an empty HyperLogLog; if all keys are non-existent, PFCOUNT returns 0.

The command is commonly used for analytics on large data streams where exact counting would be expensive. Because HyperLogLog provides an approximate count with a standard error around 0.81% for large counts, PFCOUNT returns an estimate rather than an exact value. The memory usage scales with the number of HyperLogLog structures but remains relatively small (a Redis HyperLogLog uses about 12 KB with the default precision per log; precision can be adjusted to trade memory for accuracy).

PFCOUNT is often used in conjunction with PFADD to insert elements into HyperLogLog structures and with PFMERGE to combine multiple HyperLogLogs into a single one before calling PFCOUNT. For exact cardinality, alternative approaches such as sets are required, but at a significantly higher memory cost.

Examples: PFADD hll1 a b c; PFCOUNT hll1 returns an estimate around 3. PFCOUNT hll1 hll2 yields the estimated union cardinality of both structures.