Set Sketches. They allow you compute the difference between two sets (for example to see if data has been replicated between two nodes) WITHOUT transmitting all the keys in one set to another.
Say you have two sets of the numbers [1, ..., 1million] all 32 bit integers, and you know one set is missing 2 random numbers. Set sketches allow you to send a "set checksum" that is only 64 BITS which allows the other party to compute those missing numbers. A naive algorithm would require 4MB of data be transferred to calculate the same thing.
Set Sketches. They allow you compute the difference between two sets (for example to see if data has been replicated between two nodes) WITHOUT transmitting all the keys in one set to another.
Say you have two sets of the numbers [1, ..., 1million] all 32 bit integers, and you know one set is missing 2 random numbers. Set sketches allow you to send a "set checksum" that is only 64 BITS which allows the other party to compute those missing numbers. A naive algorithm would require 4MB of data be transferred to calculate the same thing.
*(in particular pin sketch https://github.com/sipa/minisketch).