Hyper-converged infrastructure has to play well with virtual disaster recovery products. While HCI and virtual...
DR are complementary technologies, they stand alone in terms of what they provide; they have different support requirements and different end results.
The best element about virtualized DR in general is that it decouples the hardware from the software stack. The disaster recovery environment has to rely on the DR services functioning correctly.
The major selling point of the hyper-converged stack is that it reduces the costs of infrastructure, and the story is exactly the same for DR. Most DR applications deal with a virtual server as a set of files on disk, and then leverage the hypervisor API to bring those servers online. Therefore, they are treated in the same way as any other server. It is only when the disaster recovery environment is initiated that those virtual machine files are used.
The hyper-converged stack is designed to be supported by the hardware vendor, which, if required, gets the hypervisor vendor involved. Occasionally, there may be a subtle bug that needs the HCI hardware vendor and software vendor to interact and engage with one another.
Proceed with caution
In the DR scenario, HCI adds a layer of complexity. The new, secret sauce that provides massive in-line deduplication functionality at lightning speeds often takes some form of shortcut, and this can have consequences. Although rare, this can cause the machines in the disaster recovery environment to become corrupted and, therefore, unusable if certain I/O sequences occur. Such problems should be brought to the surface with the proper DR checks. All DR servers should be checked frequently.
When looking at HCI providers and DR, the obvious question should be asked: Is the HCI officially supported by the vendor? The hyper-converged stack is designed so that there is one vendor to deal with the problem end to end. Introducing software that works at such a low level needs to be thoroughly tested and certified by the vendor. The last thing an administrator wants is for a VM to fail in a true DR situation.
Hyper-converged infrastructure should work without issue. Most of the setup is essentially normal compute and a system shim that sits on top of the hosts to perform the data migration and replication between sites. Your first action should be to check with the HCI vendor on the length and breadth of support it provides for the chosen disaster recovery environment before any purchase, or worse, any real DR scenario.
HCI vendors offer disaster recovery features
Three considerations for hyper-converged DR plans
Backup and recovery emerge as HCI use cases