If the Primary site fails, the Secondary needs to take over. However, the Secondary may be behind the Primary. That is, there may be some writes that are completed to the application but that have not reached the Secondary data volumes; these writes are stored in the SRL on the bunker.
To recover from a disaster on the Primary, you can use the SRL on the bunker to update the Secondary. You activate the bunker, which converts it to the Primary role, and then the bunker can replay the pending writes from the bunker SRL to the Secondary.
The procedure is similar whether the bunker setup uses the IP protocol or uses direct storage. However, if the bunker setup uses direct storage, you must first import the disk group containing the bunker SRL onto the bunker host and recover the disk group. In both cases, you activate the bunker to connect the bunker host to the Secondary, and then replay the SRL to the Secondary.
Click the thumbnail above to view full-sized image.
After all of the pending writes are transferred to the Secondary, the Secondary is as up-to-date as the Primary. In normal circumstances, if the entire SRL is replayed, the Secondary can take over the role of the Primary with no data loss. However, in certain failure conditions, the bunker might not be able to bring the Secondary exactly up-to-date with the Primary. For example, if the RLINK between the Primary and the bunker was disconnected prior to the failure at the Primary.
Bunker replication enables you to balance the Recovery Point Objective (RPO) with the Recovery Time Objective (RTO) depending on your needs. In the case of a disaster, completely replaying the bunker SRL to the Secondary provides zero RPO. However, the RTO depends on the time required to replicate pending writes from the bunker SRL to the Secondary site. If the Secondary is far behind the Primary at the time of the disaster, the RTO could be large.
Using bunker replication, you can stop the replay after a period of time to recover as much data as possible within a target RTO. For example, if your Secondary is 2 hours behind the Primary, you can replay the full bunker SRL to achieve zero RPO but the RTO would then be about 2 hours. If your target RTO is 1 hour, you could begin bunker replay and then stop the replay after 1 hour.
If you need the application to be immediately available (RTO is zero), you can perform a normal Secondary takeover, without replaying the bunker at all. However, in this case, the pending writes in the bunker SRL are lost. In order to use the bunker SRL to update the Secondary, you must replay the bunker before performing takeover on the Secondary.
Note The bunker can act in the Secondary role, to receive updates from the Primary, or it can act in the Primary role, to send updates to the Secondary during replay. However, it cannot perform both roles at the same time and therefore does not serve as a relay between a Primary and another Secondary.
After the Secondary has been updated (either the bunker replay has completed or the target RTO is reached and the bunker replay has been stopped), the Secondary can take over the role of the original Primary. The bunker for the original Primary cannot be used as a bunker for the original Secondary. Therefore, another suitable host near the new Primary can be configured as a bunker for the new Primary.