These applications generate widely-varying workloads. Some are relatively stable, but others show periodic variation, seasonality, or anomalous events.

When workloads are stable, service providers are able to employ traditional capacity planning and make efficient use of their private capital equipment in the Data Centers. However, for varying workloads, this virtual infrastructure is sized for the worst case. This does not make efficient use of capital investment. During non-peak periods, these workloads are effectively running on over-provisioned virtual infrastructure.

Service providers see the benefits of migrating selected applications to public cloud providers, taking advantage of elastic computing and other resources. With this approach, private infrastructure can be resized for the “normal” case, with cloud computing resources handling extra load up to peaks. Applications may also be completely migrated to the cloud and then scaled up and down as needed. This paper discusses best practices for assuring cloud migrations and how to overcome related challenges.