GNU\Linux clustreing , Which level? Which application?

I need to start GNU\* clustering, But I have problem, Which level: - DB level? (replication) - file system? (distributed) - process level ? (such as intel fortran for civilization softwares) - new process ? (such as above with some differnces) I know DB level have per db backend seperated rrplication. Distrinuted file systems in linux is btrFS and freebsd uses zfs.But i have serious problem with process clustering and its softwares. Anyway , Question is, What's kernel level of process level?If I strt it , Do I have to forced to use a distributed fs?