HP 3PAR Remote Copy Software - Manual (Page 47)

2.31 Use of Virtual Copy Snapshots
InForm OS Version 2.3.1 Remote Copy User’s Guide
2.11 Use of Virtual Copy Snapshots
Remote Copy uses virtual copy snapshots (point-in-time virtual copies) of a virtual volume to
minimize the amount of data that needs to be sent over the network to resynchronize volumes
that were previously synchronized. There are several scenarios where snapshots are used.
2.11.1 In Synchronous Mode
In synchronous mode, a snapshot is created only under error or recovery situations or when a
group is manually stopped. If the backup storage server fails, or all communication links to the
backup server fail, the primary storage server stops the replication of all volume groups. It also
takes snapshots of all volumes that were completely synchronized. If a volume was still
undergoing the initial full synchronization when the failure happened, a snapshot is not taken
of that volume. That volume fully synchronizes when Remote Copy is restarted.
When the backup storage server comes back up, Remote Copy must be manually restarted
using the startrcopygroup command for all volume groups, unless the auto_recover
policy is in use (see Setting Remote Copy Volume Group Policies on page 4.4). When restarted,
Remote Copy first looks for a valid resynchronization snapshot for a volume. If the
resynchronization snapshot exists, Remote Copy resynchronizes the secondary volume by
sending only the differences between that snapshot and the current data in the primary base
volume. But before this resynchronization is started, the system takes snapshots of all the
secondary volumes that were previously synchronized. While the resynchronization is taking
place, the state of the secondary volume becomes syncing . During that time, the secondary
volume is not in a consistent state because the updates are written by location order rather
than by time order. When the resynchronization completes, the snapshots on the primary and
the backup servers are deleted.
If the primary server fails during the resynchronization, the secondary base volumes are left in
an inconsistent state, but the snapshots of the secondary volumes are left behind. When the
primary server comes back, the next resynchronization brings the secondary volumes back in
synchronization with the primary volumes.
NOTE: See Remote Copy and Thin Provisioning on page 2.33 for important
information about space allocation and deleted snapshots.