01 Mar Disaster Recovery to the Cloud How hyperconverged infrastructure can smooth the process of getting there.
The current IT goal for many organizations is “something… something… cloud computing.” The higher up the org chart one travels, the less specific anyone is on the details, but everyone agrees that cloud is important. How to get there is left as an exercise for systems administrators, and disaster recovery (DR) is a favorite first target.
Functional DR isn’t simple, even when the target environment is entirely under the control of the organization. At a minimum, for DR to the cloud to work properly, a workload’s networking has to be flexible. Hyperconverged infrastructure (HCI) can help here in a number of different ways.
HCI DR Comes In Flavors
Some HCI vendors offer cloud-based replication targets for DR purposes. This offers very similar networking environments on both ends, reducing the amount of work to be done to get DR working. These vendor-supplied DR targets can take a few different forms.
Most HCI solutions have some form of snapshot-based replication built in. The basic version of this has the production HCI cluster taking a snapshot of configured workloads on an administrator-defined schedule and then sending those snapshots to a second cluster.
The target replication cluster may or may not have the ability to light up workloads if the production cluster goes down. If the replication cluster does have the ability to run workloads, that cluster may require manual intervention to engage workloads in the event of the failure of the production cluster, or it may be a fully automated affair.
At a minimum, for DR to the cloud to work properly, a workload’s networking has to be flexible.
The most basic form of vendor-offered cloud-based DR is to simply host backup clusters in a public-facing datacenter and make them available to customers. This takes the technology originally designed for customers who had multiple sites and turns it into a cloud DR solution. The benefit to this approach is that the customer doesn’t have to pay for their own DR site. The downside is that the customer still has to pay for (or at least rent) an entire replication cluster.
Another version of this involves the HCI vendor creating virtualized instances of their HCI software. This lets the HCI vendor create (or rent) a more efficient underlying infrastructure and offer a “native” DR solution to their customers that can grow on an as-needed basis. This has the advantage of a much lower floor cost, and doesn’t have the sawtooth upgrade pattern for adding capacity that having to buy or rent physical nodes does.
Speaking Public Cloud
Some HCI vendors incorporate virtual machine (VM) translation solutions into their DR offerings, allowing VMs to be backed up directly to a public cloud provider. This is reasonably simple when sending VMs from a Hyper-V-based HCI cluster to Azure, or from a KVM-based HCI cluster to GCP. Sending something to Amazon Web Services (AWS) is a little trickier, unless it’s VMware to VMware on AWS.
In addition to getting the VMs formatted for the target environment, public clouds have very different networking environments than most HCI solutions. Additional effort is usually required to make sure that workloads will work properly in the destination environment.
At a minimum, the network environment in the cloud isn’t likely to be identical to the network environment on an on-premises network, and so effort must be put in. The end goal of moving workloads into the cloud requires making workloads composable so that they can be directly moved onto a cloud infrastructure.
Intermediate between a traditional workload and a fully composable workload suitable for the hybrid multi-cloud future, however, is simply getting DR to the cloud to work. Most HCI solutions offer an easy-button approach, but not all. Talking with vendors in depth about the different approaches available is highly recommended.
DR to the cloud capabilities are an excellent litmus test for a vendor’s planned support of cloud integration. If the vendor’s DR suggestion is to use a third-party solution to do backups and DR, ask them about their roadmap for infrastructure automation and composable infrastructure solution support. The end goal is the cloud, and it’s worth taking the time to pick an HCI vendor who is offering support for the key technologies that enable the transition.