PROTECT STGPOOL and REPLICATE NODE recommended order of operations.

Answer

The purpose of REPLICATE NODE as a stand alone command is the provide a means of making a copy (replica) of storage objects (client files, directories, ACLs and other metadata) from a source server to a target server. This is a subtle difference from PROTECT STGPOOL in that REPLICATE NODE is interested in protecting whole client files (the extents that form that file as well as the metadata that describes that file) and is not specifically tied to container pools like PROTECT STGPOOL. It is perfectly possible to replicate objects from a source container pool to a target traditional VOLUME based storage pool (such as tape)


When used together, PROTECT STGPOOL and REPLICATE NODE provide a complete protection solution when the target storagepool is sharing the same directory-container type pool. Performance testing has shown clearly that running PROTECT STGPOOL first and allowing it to complete prior to starting REPLICATE NODE will yield the best results. The reason for this is in how each of these processes views the data in the pool. REPLICATE NODE is interested in protecting whole files and processes the pool in such a way to accomplish that task. As a result reads into the pool tend to be more random and additional database activity is needed to process the metadata that needs to also be transmitted to the target server. PROTECT STGPOOL does not concern its self with client files as such. It processes only newly created extents and generates a list of deleted extents to send to the target server for future expiration. As such there is less database overhead and the reads tend to be more sequential in nature.

Because PROTECT STPOOOL handles the movement of the extents in the pool it acts as a sort of 'primer' for the REPLICATE NODE process. REPLICATE NODE no longer needs to move the extents (which is the large bulk of any replication task) and only needs to process the object metadata making even larger replication jobs run quickly.