DataCore SANsymphony-V provides many data protection features such as synchronous mirroring, asynchronous mirroring, snapshots, continuous data protection, and disk pool mirroring. In this article I will focus on the differences between snapshots and continuous data protection (CDP).
Snapshots are individual point-in-time versions or states of a virtual disk. There are two types of snapshots, Full and Differential. A full snapshot provides a full end-to-end copy of the entire source virtual volume to a destination virtual volume. This type of snapshot takes time to complete since it has to copy all the data representative of the source virtual volume.
In contrast, the differential snapshot ‘appears’ to be a full copy of the source volume, except no data is copied upon enabling the snapshot. The destination volume, which represents the snapshot of the source volume, is initially blank. So here with where the magic happens… so pay attention. Before inbound write data is committed on the source volume, a copy of the source block that is about to be modified is copied to the destination snapshot volume. Then the source commits the inbound write data modification. This method allows the destination snapshot to always represent the original state of the source at the time the snapshot was taken and it doesn’t interfere with the source volume data in any way. This is also referred to as a copy-on-write snapshot operation.
Differential snapshot operation detail:
Differential snapshots are excellent for fast live copies of production volumes and for eliminating the so-called backup-window. Snapshots also work hand-in-hand with VSS when OS or application-level quiescing is necessary. Some other interesting options that are available once the snapshot is in place are:
- Update Snapshot (which updates the destination with the changes that have occurred on the source volume)
- Revert From Snapshot (which returns the source back to the original state from when the snapshot was taken)
- Promote To Full (which turns the differential snapshot into a full snapshot maintaining the state of the original snapshot)
The last option I want to point out in this section is a very useful capability called: Duplicate Disk Identifier. This option is very helpful when working with VMware ESX servers. Before this option existed, if you ever presented a snapshot volume or a replicated volume to an ESX host, first you would need to enable Disk Resignaturing on the host, then rescan the HBAs on the host, then once the volume (Datastore in ESX-speak) appeared, rename the volume name from ‘snap-8a2d0912-OriginalName’ to ‘OriginalName’, register the VM’s, and then start them up…. Whew! Now with this Duplicate Disk Identifier option, you no longer need to enable resignaturing or rename the datastore once it appears. Assuming that you created the snapshot with this option enabled, now you simply map the snapshot, rescan the host HBAs, register the VMs and go.
So if you thought all of this was pretty cool, you ain’t seen nothing yet!
SANsymphony-V Continuous Data Protection
If snapshots are individual point-in-time versions or states of a virtual disk, then CDP represents period-in-time versions or successive states of a virtual disk. CDP is essentially the tape roll that records every single write operation that has occurred on a volume since CDP protection was enabled.
CDP operation detail:
So imagine that you have a particular virtual volume mapped to an application server. This volume has been under CDP protection for the last 10 days. Then today at 09:37:46 AM a virus destroys the MFT on the volume and all the files are now inaccessible. You get the call that something bad has happened. After some investigation, you determine that the volume containing the data is completely unrecoverable from the OS and file system perspective. So now you turn to CDP to save the day.
You open the SANsymphony-V console, you select Create Rollback, you select a recovery point (specific time when you want the volume restored to) – in this case let’s say 09:37:40 AM (so that the maximum potential data loss is reduced to six seconds), and click Create. Now that the recovery point has been established, you simply serve the virtual volume restore point to an application server. DONE! Also worth noting here, the Duplicate Disk Identifier option available with Snapshot is also available with CDP. This is a checkbox on the Create Rollback screen.
Although not required, CDP can also be used with VSS. Simply call VSS any time you choose on the host, making note of when the VSS quiescing occurs. Then when you rollback, you know exactly when you have a flushed and consistent state on the disk.
DataCore’s Snapshot and CDP features are very elegant and very powerful. They are similar in that they protect the virtual volume state, but they achieve it in different ways and there are different use cases for each. Clearly, CDP is a more advanced form of snapshot and certainly offers the highest level of survivability in the event of data corruption or loss.
For more information about Snapshot and CDP in SANsymphony-V, please visit DataCore Software.