Restarting DFS after an abnormal system shutdown is faster if DCE LFS is used because DCE LFS logs information about operations that affect the metadata associated with DCE LFS aggregates and filesets. When the system is restarted, DCE LFS replays the log to reconstruct the metadata. It returns the system to a consistent state faster than non-LFS file systems that must run the fsck command.
Access to information is more reliable in DFS for a number of reasons (in addition to the logged metadata already mentioned). In a distributed file system, multiple clients such as the Cache Manager can attempt to access the same data simultaneously. DFS uses tokens to ensure that users are always working with the most-recent copy of a file and to track who is currently working with the file. Tokens identify operations the client can perform on the data. They also act as a promise from the File Exporter that it will notify the client if the centrally stored copy of the data changes; following such notification, the client can then retrieve the most-recent copy of the data the next time it is requested by a user.
DFS also improves the reliability of data access by allowing you to replicate commonly used DCE LFS filesets on multiple File Server machines. When you replicate a fileset, you place an identical copy of the fileset on a different File Server machine. The unavailability of a single server that houses the fileset generally does not interrupt work involving that fileset because the fileset is still available from other machines.