When part of the file system suffers corruption (detected either by checksum errors, safety checks in the file system code, or errors reported by the underlying storage), only the chunk containing it must be checked. Likewise, when the system crashes, only chunks with metadata being actively modified at the time of the crash must be checked before the file system can be brought back online. Since the chunks are relatively small (order of a few gigabytes), the fsck time is correspondingly short.
