A method and system are provided for performing an extract-transform-load (ETL) process. The method includes collecting load information about a volume and a complexity of raw data to be processed during the ETL process. The method further includes receiving an expected completion time of the ETL process and execution information about (i) hardware resources and (ii) an influence of the hardware resource on an execution time of the ETL process. The method also includes calculating resources for a distributed processing software infrastructure to be used to perform the ETL process, by applying a statistical method to the load information, expected completion time, and execution information. The method additionally includes dynamically assigning cloud resources corresponding to and based on the calculated resources, in accordance with the expected completion time. The method further includes performing the ETL process on the raw data using the assigned cloud resources and storing ETL process results.