A customer recently asked about how to restore DFS Replication replicated folders after a bare metal restore. Shobana Balakrishnan provided the following answer:

DFS Replication does not depend on the cached configuration (xml files) and obtains its configuration from Active Directory. If you have set some machine-specific parameters, such as RPC port, you may need to re-issue the command via WMI.

For bare metal recovery, the only tested and recommended approaches we have for R2 are:

(1)     To do an ASR restore of the entire system – all volumes.  In this case, the data is restored without the knowledge of DFS Replication (i.e., before services are started up), and DFS Replication on startup (assuming the AD configuration is the same) will poll AD and treat all replicated folders that it subscribes to as new.  The folder will go through initial sync with partner, and if the files restored are identical, they will not be replicated over the wire.  If different (that is, files on partner are more recent), RDC will be used to replicate only the changed chunks.  Files that are on the restored machine but deleted on the partner will be moved to Preexisting.  It is important to note that DFS Replication does not know this is a restore and hence will respect connection schedule.

(2)     To do an ASR restore of the system volumes and a component restore of the replicated folder when DFS Replication is running.  When DFS Replication starts up with no restored data, it will start initial synchronization with its partner and start replicating the data over the wire.  Note you can disable the connection if you do not want this to happen.  If you then perform a use-selectable restore of the replicated folder(s) using a backup/restore app, then DFS Replication will detect that a restore has been performed and will perform internally a recovery sync, which is similar to initial sync except that the connection schedule is ignored.  Otherwise, the semantics are the same – if files are different (that is, files on partner are more recent) RDC will be used to replicate only the changed chunks.  Files that are on the restored machine but deleted on the partner will be moved to Preexisting.

One pain point for this process is the time it takes to complete initial sync over a throttled link.  With approach (2), recovery should go faster since the schedule is ignored.  Note that < 1KB per file/folder should actually replicate across the network, so your bandwidth consumption should not spike

 --Shobana