Home > Capacity Planning Projects (this page)

Capacity Planning's primary role is to forecast workload demand over some future timeframe and the corresponding service levels that will be provided at the maximum forecast workload level. If the projected service level is insufficient to meet the applicable service-level requirements, Capacity Planning is charged with recommending configuration changes that will provide adequate performance at the demanded throughput level.

Similar to performance management your service level requirements and objectives are important to capacity planning. Utilization is typically the first measurement discussed with capacity planning, but many other metrics are also required to monitor and manage the system's capacity. These include:

Utilization

Measures of occupancy of tangible resources, such as control units, devices, memory, CPU; by workloads.

Load

Measures of the utilization of resources or services such as jobs, transactions, CPU hours, I/Os, etc.

Service

Response time, throughput, service time, percents within targets, etc.

System Behavior

Measures of internal logical behavior such as queues lengths, and paging.

Latent Demand

As any system approaches saturation, existing load may be displaced from a selected processing window and new loads may be deferred from the system. These deferred loads can have a profound effect on a system after the bottleneck is removed. Consider the paradox of a 25% upgrade... we were 100% busy before the upgrade and 120% after... where did the upgrade go? You may have experienced this first hand.

Last Updated: 8 September, 2003