Figuring out which disaster recovery solution is best for you is likely to involve different criteria. Hard metrics that are typically quoted are RTO (recovery time objective) and RPO (recovery point objective). You’ll often see them used in service level agreements for data recovery for instance. However, while being a good start, these two well-known parameters may not be sufficient. For example, to recover just one crucial piece of data, you may need to recover all of your data, which may be a long time indeed. Additional metrics may therefore provide a more accurate picture of whether or not a solution will suit you or your organisation.
In terms of hard metrics, i.e. those that yield quantifiable, directly comparable information, it is possible to extend to three groups – recovery time characteristics, recovered data characteristics and recovery scalability characteristics. RTO is an old friend in the first group, and now accompanied by RTG (recovery time granularity). RTG defines recovery point options with regard to logical failures. This is the subtle difference between RTG and RPO (recovery point prior to a physical failure). RPO is then in the second group, together with ROG (recovery object granularity for the level of individual objects than can be recovered), REG (recovery event granularity for recovery from specific events).
Further criteria cover usability of data by an application, geographical scope within which protected data must be held ready, scalability over different numbers of applications, resiliency of the recovery solution itself and cost efficiency. This last metric takes into account how much system administrator effort is required to use the solution, as well as the amount of IT resources needed to implement it. It underlines another important point. The disaster recovery solution you choose must also be one with which you feel comfortable and that you can easily apply even when panic is all around you – a key point to remember at the same time as making measurements and comparing numbers.