Optimize a file system for different types of applications in a data processing cluster using dynamic block size granularity

Embodiments of the invention relate to the optimization of a file system for different types of applications in a data processing cluster using dynamic block size granularity. An exemplary embodiment comprises reserving a predefined number of memory allocation areas for each node in a cluster, each memory allocation area comprising a set of contiguous blocks of data on a single disk of a locally connected memory; Use of a contiguous set of data blocks on a single disk of a locally connected memory as a block mapping granularity for striping of data to be stored for a data processing operation in the cluster in the file system; and using a single data block of a shared storage system as a block allocation granularity for striping of data to be stored for a data access operation in the cluster in the file system.