DCE LFS allows you to replicate (copy) DCE LFS filesets. When you replicate a DCE LFS fileset, you place read-only copies of it on multiple server machines. The unavailability of a single server machine housing a replicated fileset does not usually interrupt work involving that fileset because copies of the fileset are still available from other machines. The replication of commonly used configuration and binary files on multiple server machines greatly reduces the chances of their being unavailable as the result of server machine outages. Replication also prevents a machine from becoming overburdened with requests for files from a frequently accessed DCE LFS fileset. Replication is supported only for DCE LFS filesets, not for non-LFS filesets (file systems on non-LFS partitions).
Two types of replication are available with DCE LFS: Release Replication and Scheduled Replication. With Release Replication, you issue a command to copy a source fileset to the server machines housing its read-only replicas every time you want to update the replicas to reflect the current contents of the read/write fileset. This type of replication is useful if the fileset seldom changes or if you need to closely monitor the replication process.
With Scheduled Replication, you specify replication parameters that dictate how often DFS is to automatically update replicated filesets with new versions of source filesets. This type of replication is useful if you prefer to automate the process and do not need to track exactly when releases are made. Both types of replication produce the same result: source filesets are copied to different server machines. The system administrator chooses which type of replication to use with each fileset.
Note: Replicas of sparse files do not expand to their full size. Replicas have a minimum granularity of 64 KB; any 64 KB chunks that do not contain actual data require no storage space. Chunks that do contain data expand to occupy a full 64 KB of storage space.