As the cost of storage decreases more and more data is kept online, including most recent backups of that data. In the event data is corrupted or deleted, having a recent online backup allows it to be quickly recovered using FlashCopy. An added benefit is that online backup copies can be written to the offline tier, tape or object storage, before the next FlashCopy backup cycle without impacting applications. FlashCopy is also widely used to quickly copy or move of datasets. But up until now, FlashCopy has not been compatible with some replication topologies that include Global Mirror.
The DS8000 and DFSMS team shipped the first phase of FlashCopy onto Global Mirror primaries support. In the initial support, the Establish FlashCopy command is executed at the primary, and the FlashCopy data is transferred over the Peer to Peer Remote Copy (PPRC) links. Note: Since the data is drained over the PPRC link, FlashCopies that are done frequently or include a large number of tracks could elongate the Recovery Point Objective (RPO).
If you have a pool of storage dedicated for FlashCopy backups that is managed outside of your existing Global Mirror topology, it makes sense to continue to manage those FlashCopy backup volumes as you do today. The initial support will allow you to use FlashCopy, instead of a slow copy method, to recover data sets back into your GM topology, improving application availability. You will still have to consider the RPO increase, but given recoveries like this should be infrequent, it may be acceptable in your shop.
This solution will use the continuous delivery model and the initial support is available on z/OS V2R2 and above along with DS8880 Release 8.5 Service Pack 4.
For more information:
- DS8880 Knowledge Center
- APAR OA57173