Software as a ServiceSoftware as a service may offer customers an SLA whereby the service is expected to scale to handle peak load without slowing down. The service may be managed to automatically scale when resource demand hits a preset level. This is hidden from customer who simply sees that the service works at a reasonable speed.
Cloud APIsCloud APIs include interfaces for requesting more resources. Services designed for cloud may request resources based on events, resource utilization, management decisions and other application specific criteria.
Infrastructure ManagementThe providers of cloud infrastructure may automatically power down machines that aren't being used. A pool of running unallocated servers is maintained based on demand forecasts or actual observed demand.
Scheduled ScalingA bank schedules more resources during peak banking hours for backend processing such as settlement.
Predictive ScalingA video-on-demand company uses predictive analytics to request resources to handle anticipated demand based on historical patterns. If people watch more movies on Valentine's Day, the service will scale up more servers that evening in preparation. Such analytics could potentially consider a large number of factors such as the weather.
A feature of cloud APIs that allows resources to be automatically scaled up and down based on a schedule or application specific factors such as events.
Conserving resources such as electricity.Reducing cloud service charges such as CPU hours.Responding to peaks in demand for a service.