The Capacity Manager should carry out monitoring of capacity in terms of utilization monitoring, response time monitoring, threshold reviews, etc. This will help to obtain an insight into working patterns and peaks and troughs of the technology capacity.

Few examples of data that can be monitored include:

• CPU utilisation
• Memory utilisation
• Per cent processor per transaction type
• IO rates and device utilisation
• Queue lengths
• Disk utilisation
• Transaction rates
• Response times
• Batch duration
• Database usage
• Index usage
• Hit rates
• Concurrent user numbers
• Network traffic rates.

The data collected should be analyzed to identify trends, based on which the normal utilization and service level baselines can be established. The Capacity Manager should regularly monitor and compare data against these baselines and identify and define exception conditions in the utilization of individual components. It also helps to identify issues such as bottlenecks, inappropriate distribution of workload across available resources, breaches or near misses in the SLA; report them and take corrective actions on the same. The key purpose is to forecast issues, wherever possible, by monitoring changes in performance and its impact. Data can also be used to predict future resource usage, or to monitor actual growth against predicted growth. Technical infrastructure should be reviewed to identify components to be used, replaced or exchanged to meet the technology availability targets.

The Capacity Manager must also collate data on the performance and capacity of IT product components, and carry out application sizing, trending and modelling. This should be done to forecast future capacity levels and identify gaps (if any) with respect to future capacity requirements.

Modelling is a technique used to forecast capacity and its performance, and also assess the impact of changes to the technology capacity components. This is done by using models for assessing different scenarios and recommending appropriate actions to ensure business requirements are met.

Application sizing evaluates the necessary resources to run new or changed applications. The resulting predictions include information about expected performance levels, necessary hardware and associated costs.