Although hyper-converged infrastructure systems were originally introduced as a standardized platform for running virtualized workloads, their uses are beginning to evolve.
One element that sets hyper-converged infrastructure systems apart from other types of server appliances is that they almost always include an integrated storage array that can be accessed by all of the nodes within the hyper-converged appliance. In many cases, this storage array includes both flash and traditional storage, thereby allowing for the creation of storage tiers.
Because hyper-converged systems were originally designed to act as virtualization platforms, the assumption was that the compute nodes within the hyper-converged appliance would act as failover cluster nodes, and the storage would be configured as a cluster shared volume, thereby providing high availability to the virtual machines running within the cluster. Over time, however, hyper-converged infrastructure vendors began to realize that hyper-convergence was ideally suited to aid the backup and disaster recovery process.
Hyper-convergence gets hyped for instant recovery
A hyper-converged appliance's integrated storage can be repurposed for use as a backup target, but that is not the only reason why hyper-convergence is so ideally suited to the backup and disaster recovery process. The bigger reason is the hardware's ability to perform instant recovery.
Disk-based backup systems, including hyper-converged platforms, are often able to perform instant recovery for virtual machines. In order to do so, however, the backup platform must have sufficient compute, storage, network and memory resources to allow it to host a copy of the virtual machine.
Each vendor handles instant recovery in its own way. Generally speaking, however, a recovery point within the backup target is synthetically converted into a virtual hard disk. A virtual machine instance is then created directly on the backup server and attached to the newly formed virtual hard disk. Before the virtual machine is brought online, the backup software creates a snapshot of the virtual hard disk. This snapshot is essentially an empty virtual hard disk, and it exists for the purpose of protecting the backup contents against modification. When the virtual machine is run on the backup server, all write operations are directed to the virtual hard disk that is associated with the snapshot, meaning that the backup contents can be read, but are never at risk of being overwritten.
With the virtual machine now running on the backup server, the administrator can begin working to fix the problem that prompted the need for the disaster recovery process. Once the problem is fixed, the administrator can initiate a traditional restoration of the virtual machine, while the VM copy residing on the backup server continues to run. When the restoration is complete, the contents of the snapshot are merged into the newly restored VM. At that point, the VM is up to date and can resume normal operations.
When it comes to instant recovery capabilities, it is important to understand that nearly all of the major backup vendors offer some sort of virtualization-based instant recovery feature. Because these vendors have implemented instant recovery at the software level, it is not unique to hyper-converged environments. All an organization really needs if it wants to perform instant recovery is a backup application that supports it and a compatible, disk-based backup target with enough compute, memory, network and storage resources to make it practical. Even so, hyper-convergence works especially well for this type of disaster recovery process, because there are vendors that offer integrated hardware and software products.
Cohesity hits the clouds
Hyper-converged backups continue to evolve, and one of the latest trends is public cloud integration. Cohesity, for example, offers a purpose-built, on-premises, hyper-converged backup product that the company refers to as a Cohesity Cluster. In addition, the company has recently begun offering a public cloud marketplace platform that allows customers to create a virtual Cohesity Cluster in the cloud.
Initially, cloud-based Cohesity Clusters served two purposes. They could back up other cloud resources, and they could act as replicas of on-premises Cohesity Clusters. So a cloud-based Cohesity Cluster can act as an off-site backup replica and can also be used for cloud-based disaster recovery. More recently, Cohesity has begun giving its customers the option of backing up on-premises resources directly to a cloud-based Cohesity Cluster, without the need for an on-premises one.
Although Cohesity is one of the best-known hyper-converged backup vendors, it is not the only option. Other vendors offering hyper-converged backup and disaster recovery products include Rubrik and Nutanix.
Hyper-converged DR finding more users
One of the reasons why cloud-based hyper-converged backup and recovery systems are so significant is they make the technology accessible to smaller organizations. Although hyper-converged systems were initially marketed as an inexpensive alternative to traditional servers, they often cost tens of thousands of dollars, with higher-end units costing in the six-figure range. As such, hyper-converged systems have been historically favored by small enterprise environments and by large enterprises that need to place computing resources at branch offices. Cloud-based hyper-convergence opens the technology for use by much smaller organizations.
Hyper-converged systems are probably best known for their use as platforms for hosting virtual servers or virtual desktops, but hyper-convergence can be an ideal platform for the backup and disaster recovery process.
HCI vendors perform disaster recovery in varied ways
Options for hyper-converged disaster recovery
Virtual DR planning can be strong and cost-effective