Technology-Webedition

The D test as part of disaster forecasting
A disaster is said to have occurred when business‐critical functions of an IT system are no longer available. To completely rule out this kind of scenario, D tests are performed and monitored based on existing disaster recovery plans (disaster testing). There are two major focuses ... [more - pdf]

To test the resilience of the datacenter supply structure against power, airconditioning, ventilation system, or Internet connectivity failure, tests are arranged to, e.g., deliberately interrupt the power supply. To pass the D test, the system landscape (servers, storage systems, network components, etc.) need to stay up without interruption.
The second test checks whether it is possible to restore inconsistent or corrupted data from backups.

How does cloud architecture change this?

In the case of a private cloud the same preconditions apply to D tests as for any other "legacy" datacenters. In the case of a public cloud the question is whether you can even perform a D test. Testing the continuity of a datacenter supply structure can turn out to be difficult in a public cloud; one can't perform a D test at an arbitrary point in time. Because various services are spread over multiple datacenters in multiple countries, a holistic D test would need to cover all of the resources and services involved. Because it is impossible to tell which resources or services are provided by which datacenter (and differences due to time zones must also be taken into account!), legacy D tests are no longer a viable option. Testing whether you can restore data from backups is the same as for a "local datacenter"“; cloud architecture may need careful planning, but it is similar to legacy procedures in terms of backup/restore.

Chance Cloud

The previous section showed that D tests relating to the datacenter supply structure are virtually impossible to perform in a public cloud. Despite this, you can assume that this isn't a disadvantage. The cloud in particular offers the ability to compensate for failures of technical components thanks to its ability to bring in resources from other datacenters. This means that the topic of disaster testing in the datacenter supply structure for a cloud might be unnecessary.