Cdp; Consistency Groups; Consistency Group Rollback - American Megatrends StorTrends 1300 User Manual

CDP

Continuous Data Protection. AMI definition of CDP is a method of uninterrupted
safeguarding against data loss or corruption, by using one of a number of possible
services or methodologies. Generally speaking, CDP services operate by continuously
capturing all changes to data and storing these changes in a separate storage location.
CDP services also provide the capability for virtually instantaneous restoration of data to
any point in time prior to a catastrophe or data loss.

Consistency Groups

The concept of consistency groups on Snap Assisted Replication provides a mechanism
to group application dependent volumes. Application aware consistent snapshots can be
scheduled (for example, through VSS for SAN volumes and Snap scheduler for NAS
volumes) to be taken at the same time on the dependent volumes, thereby ensuring
consistent recovery points in the primary. When such volumes are grouped through
Consistency Groups while defining Snap Assisted Replication, replication will also
happen across consistent snapshots. Snap Assisted Replication module will ensure that it
completes replicating all snaps in a particular snap group, before it starts with the next
snapshot. No snapshot taken at a later time T1 will be replicated, until the prior snapshot
taken at time T0 on all volumes are replicated and snapshots for time T0 on all volumes
are replicated in the secondary. Similarly, on a failover to secondary, all volumes in the
consistency group are rolled back to the same snap group there by maintaining
application consistency across all the volumes.

Consistency Group Rollback

In the event of an unforeseen incident, the administrator might be required to rollback the
volumes in the consistency group. By nature of Consistency group definition, rollback
can happen only across all volumes and NOT alone on individual volumes. There are two
scenarios during a consistency group rollback. Rollback happens to a snap group at time
T(x+n), while replication is currently happening on snap group at time T(x), where
T(x+n) > T(x) – rollback is happening to future snapshot than what is currently replicated.
This scenario will NOT affect replication. Replication will continue to happen without
any impact as the rollback is not happening around the replication snap group time. In
another scenario, rollback might happen to a snap group at time T(x-n) while replication
is currently happening on snap group at time T(x) – rollback is happening to a past
snapshot that what is currently replicated. This indicates that the disaster is also replicated
to the remote and hence, the remote is also affected. The only solution is to rollback the
remote as well to the consistent point T(x-n). So, both ACTIVE and PASSIVE systems
will rollback to the snap group at time T(x-n) and replication will start from there.
Glossary 315