A storage system according to the present invention comprises a controller and multiple storage devices, constitutes a RAID group from (n+m) number of storage devices, wherein the storage system stores write data from a host computer to n number of storage devices, and stores redundant data generated from the n number of data to m number of storage devices. When failure occurs to at least one storage device, the controller reads data in a compressed state and redundant data from each of the storage devices where failure has not occurred in the storage devices constituting the RAID group, and transmits the read data in the compressed state to the storage device which is a data recovery destination.