A universal distributed model for password cracking

Abstract

Due to the parallel nature of brute force attack, it is well suited to use parallel or distributed computing to recover passwords encrypted by one-way hash functions. When distributing the password cracking task between threads or nodes, it is necessary to divide the search space into several subspaces. It is ideal that the size of these subspaces is the same for a load balanced solution. However, increasing the length of passwords will raise the number of combinations exponentially. This will raises issues for the search space uniform partition. This paper presents a universal distributed model for password cracking capable of dividing the search space evenly. For this, the paper addresses the definition of a base-n-like number, and discusses operations on it and relations with base-n number. The model is data independent, easily implemented and needs a few of communications. Experimental results are then shown.