The invention provides a distributed big data function dependency discovery method based on a Spark platform, and the method comprises the steps: carrying out the partitioning of data according to thenumber of CPU kernels distributed by each node of a Spark cluster; generating all the non-empty subsets of the attribute set, including generating a set containing all the non-empty subsets through all the attribute sets in the database, preparing for solving the number of equivalent classes of all the attribute sets; accumulating the equivalent class number of each node attribute set, obtaininga (attribute set, equivalent class number) set of the global database through equivalent class calculation; Iterating each attribute set to generate a function dependency set, namely constructing a candidate function dependency relationship by the subset of each attribute set, and judging whether the function dependency relationship is established or not. According to the method, the problems of unbalanced load and low efficiency of a function dependency discovery algorithm in a distributed environment are solved, and the execution efficiency of function dependency discovery is greatly improved.