When the new release of v1.70 is complete the function to provide Target Volume Replication is very similar to RAID 1.
The first stage is rebuilding of the destination to be the copy of
the source volume. Next stage (endless) is online modifying the destination
volume to keep the mirror. The difference comparing to the RAID 1 is that,
after any broken connection only the changed data will be 'rebuild' (again
first stage takes place) on the destination.
Note: the replication should be performed on separate NIC and the
connection should be as fast as the main network connection, to prevent
lowering data throughput. Once you create the Target Volume you have the choice to set it as Source or Destination to the mirrored server. Replication of volumes should
be same size.
Changing destination to source without loosing data is only possible when
data is consistent.
This is a manual operation on how to change source to destination:
1. Stop replication task on source or turn off server
2. Now it will be possible to change destination to source, when the data is
consistent.
Basically this is two boxes replicated via DRBD and then load balanced. Can we not just load balances the requests (reads/writes) between the two and count the on replication engine to replicate the data?
I have read thru some of the documentation and researched in our database to see if this has been performed by any of our previous customers and could not find any. Currently we have Target Volume replication with new version 1.70 on our website for this update. If adding additional replication from what I read, this site references to provide highly available NFS servers, the iSCSI Enterprise is not a file sharing server. This would require our NAS-XSR Enterprise product. You can test this by downloading NAS-XSR Enterprise for 30 days link is below.