Public cloud consumption is typically charged based on the size and type of provisioned resources, not how or even whether they are used. Optimization recommendations help ensure that cloud resources comprising a workload are provisioned to minimize cost while still providing the required performance and functionality. While it’s theoretically feasible to manually identify idle/overprovisioned resources, and determine the appropriate optimization action, in practice some sort of automated tooling that provides recommendations is normally used (with the appropriate business context such as tagging/mapping of resources), to assist with the scale and speed of resource management. Optimizations may consist in areas of resource type, size and number, and also service selection. In addition, organizations often find it challenging to actually execute or action recommendations, so consideration must be given to removing barriers.
The effort involved in implementing an optimization recommendation may vary from workload to workload. It is important to balance the effort of the initial resourcing with potential future recommendations. Workloads that are more static, have larger resource costs, and larger costs of implementation/testing/verification should have larger investments in initial resourcing, and be potentially exempted from optimization recommendations. Smaller and more agile workloads will benefit more from optimization recommendations and have this factored into the workload architecture.
It is important to ensure that the data used for the recommendations covers the workloads normal usage cycles, ensure regular increases in activity such as end of month activities are factored into the data.
Factor in future growth/usage when analyzing recommendations. If the resources will require another change in a short period of time, the initial optimization work will most likely not provide a business saving.
Factor in the time it would take to undo the optimization and potential business impacts when optimizing. The business may decide to over-provision internet facing resources, as the time to scale up in the event of a crisis, and possible brand damage may outweigh the additional costs in resourcing
The correct approach with optimization recommendations is to implement recommendations where the cost of implementing the recommendation is less than the benefit received, this provides an overall saving to the organization.
As vendors create more resource types, there will potentially be an increasing amount of recommendations. It is important not to worry about the volume of recommendations and look at value only