Hyperconverged Infrastructure (HCI) has been the new shiny for some years now and for good reason. HCI clusters servers and storage together through a hypervisor to distribute data across these clusters and nodes for resiliency as well as mobility, lowering operating and storage costs. HCI vendors, however, will claim that the solution can eliminate the need for backup because of of the potential for unlimited snapshots, giving the platform a point-in-time capability. You can, in theory, roll back a VM to almost any point in time as long as a snapshot exists for it.
What really gives HCI vendors the intestinal fortitude to claim you don't need backups is the asynchronous replication of snapshots. The platform can collate snapshots and send the delta (the change) between snapshots to a different location. Some HCI solutions can then consider the VM data protected. It's true that the dedup copy requires almost no extra space but the copy is dependent on the metadata staying intact. This creates a level of vulnerability of your systems and data for a disaster recovery scenario.
With HCI you have a solution that delivers resiliency and redundancy!
But it doesn't provide a true backup. A true backup is a secondary copy of data on a completely separate storage infrastructure running separate storage software.
The problem with HCI as a "backup" is fourfold.
First, HCI only protects data within the HCI environment itself. If your data center isn't all on an HCI platform, your systems not on it are exposed to risk of loss.
Second, HCI is wholly dependent on the storage software managed by the hypervisor. The dedup process relies completely on metadata tables of the VM replications. If the tables are corrupted or fail and there's no backup version of the table, you'll have to recreate all that data. That means no point-in-time recovery option.
Third, HCI snapshots don't have the ability to let you find specific data in the snapshot or even across them. This can be problematic if you're looking for a specific file, folder, or a version because it may be in one of hundreds of snapshots. That means you may be forced to search through them all to find a file.
Fourth, a truly redundant HCI requires you to stand up environments in your primary data center AND and a secondary disaster recovery site. If your remote DR site isn't resilient, your HCI isn't going to serve your recovery operations.
As you consider your move to HCI, you should ask your HCI vendor to address these concerns and consider the risks of not having a true backup to help you recover from a catastrophic loss.
Firestorm Cyber is a Managed Security Service provider. Our enterprise-grade Disaster Recovery as a Service (DRaaS) is a solution that provides rapid, affordable fail-over to a second site with the flexibility to boot from the appliance or cloud. Individual file and folder recovery is snap and takes minutes.
When disaster strikes, fail-over applications (VMs) to the cloud or locally in 15 minutes or less — usually much less — guaranteed!