How do snapshots work in the public cloud?
Snapshots work at the disk/volume level. Snapshots create a full copy of the volume for the first time and then incremental copies after that. All snapshots are stored in object storage in the cloud.
Cloud snapshot challenges
Following are some of the challenges associated with the cloud snapshot approach:
- DBAs care about application consistency. But unfortunately, most clouds lack application consistency for most workloads. Even the crash consistent snapshots don’t offer write order fidelity when a VM has multiple volumes.
- With cloud snapshots, search and file/folder level granular restores are not possible. Thus, all the VMs running various business modules associated with SAP application will not have granular file/folder restore capabilities.
- In some clouds, there is no policy orchestration of snapshots resulting in a lot of manual scripting process. Moreover, cataloging of snapshots and recovering from a point-in time with a VM centric view is not possible.
- Cloud snapshots use object storage but is expensive at ~$50/TB/month especially for long term data retention, as compared to storing backups in cheaper object storage tier in the cloud. For more details, CTA.
- Accessing cloud snapshots is fast. However, performance suffers as the data has to be copied from cloud object storage back to block storage. Thus HANA recoveries or provisioning HANA clones to test/dev environments will take a lot of time.
Actifio GO is a true SaaS data protection and recovery platform for SAP using HANA, Oracle, MS SQL databases running in Cloud VMs. It delivers application consistent, incremental forever backups, instant recovery for DR, and rapid DB cloning for test/dev.