Backups and Restores Are More Efficient When Deduplicated
Deduplicating all data once and forever across all stages of the data lifecycle includes backup data. When creating a local SimpliVity backup, the operation that executes is simply just a copy of the metadata of the VM; no data needs to be read, copied or written. This copy only requires a few IOPS and happens completely within SSD, so the entire operation only takes about 15 seconds to complete. In the end, you end up with a fully independent logical copy of the data, with no ties to other VMs or backups like you would for snapshot-based or incremental/differential based backups.
When restoring a backup, the same operation occurs; the only difference is that the resulting metadata copy is presented as a VM to the hypervisor through the datastore, with the additional task of registering the VM within vCenter. Total operation time to restore is usually about 20 seconds.
Given that a local backup or restore is nothing more than a metadata copy operation, no IO is created to the HDDs since no data needs to be read or written. This means there are more IOs available for other read/write operations that need to occur.
Another advantage to a SimpliVity backup is that there is no reliance on the hypervisor layer, which reduces strain on the hypervisor resources. There’s also no reliance on hypervisor backup APIs, such as the vSphere APIs for Data Protection (VADP) or VMware’s Changed Block Tracking (CBT). Both technologies have been known to have problems, especially with the reliance on vSphere snapshots, not to mention changes to the APIs over the years. An additional side benefit is that there is no special recoding of the backup or restore process needed to support different hypervisors – the same code creates backups for vSphere and KVM.
Testing a SimpliVity backup is as simple as restoring the backup with a different name (by default it will be given a new name based on the time of recovery) and isolating the network. This can be done on a VM-by-VM basis since we manage data at the VM level, unlike traditional storage array snapshots. It should take less than 30 seconds to get the power on operation started. With open APIs, and integrations into VMware vRealize Automation and Cisco UCS Director, the restore and network isolation processes can be fully automated, along with the validation testing and clean-up tasks, for a faster and more repeatable process.
By deduplicating backup data along with all other data on the system, less space will be consumed for each backup, allowing for more frequent backups and longer retention periods. This deduplication happens across all VMs and datastores, and includes all local and remote backups. At the time it is created, a backup is 100% deduplicated data with no dependencies on the original VM.
All of this combines together to result in shorter time between backups and quicker restores. In fact, we have one customer taking backups of every VM every 4 hours and retaining them for 1 year. If you’re interested in learning more about how SimpliVity improves data protection, watch our recent data protection webinar with Forrester.