ZFS is solaris file system which provides simple administration and support very large data transaction. it is a basically new way to data management. Solaris has blown away 20 years of obsolete assumptions, eliminated complexity at the source, and created a storage system that’s really a pleasure to use.

All operations in ZFS file system are copy-on-write transactions, therefore on-disk state is always valid. There is no need of fsck etc. for ZFS file system. Every block of ZFS is checksummed to avoid data corruption, and the data becomes self-healing in replication (mirrored or RAID) configurations. In case of damaged data, ZFS detects it and utilizes other copy to fix it.

ZFS introduces RAID-Z just like RAID-5 with variable stripe width to eliminate the RAID-5 write hole. RAID-Z writes are full-stripe writes. There’s no need for NVRAM in hardware hence ZFS loves cheap disks.