Capacity planning
Forecasting how much processor, memory and storage the mainframe will need, so there is enough before it is needed.
Also written capacity planner, capacity management, performance and capacity
Capacity planning makes sure the mainframe has enough resources for the work coming its way. Too little, and services slow down or batch misses its deadlines. Too much, and money is wasted, since capacity and software costs are high.
Planners start from measurements, mainly RMF and SMF data, showing how busy processors, memory and storage have been over time, especially at peaks such as month end. They combine that with what the business expects: more customers, a new application, a migration onto or off the mainframe. From this they forecast when more capacity will be needed and how much.
The answer is not always to buy more. It may be to move work to quieter times, to send eligible work to zIIPs, to tune expensive programs, or to adjust WLM goals and LPAR weights.
Capacity planning is a specialist role, often combined with performance analysis. Developers meet it when a new application must be sized before it goes live, or when a change makes a job noticeably more expensive.
Where can I learn this hands-on?
Right here, at ZedBank. You take a junior job at a simulated bank and learn each part of the work by doing it, with Zed beside you. There's nothing to install.
Start at ZedBankRelated
- MIPSMillions of instructions per second: the common, rough way of describing how much processing capacity a mainframe has.
- RMFResource Measurement Facility: the z/OS tool that measures how busy the system is and shows which work is waiting for what.
- SMFSystem Management Facilities: the z/OS feature that writes a detailed record of what happened on the system, used for billing, performance and security.
- WLMWorkload Manager: the part of z/OS that decides which work gets the processor, memory and disk access first, based on goals the site sets.
- Central processorA general purpose processor on the mainframe, the kind that runs z/OS work of any type.
- ChargebackCharging each department or application for the mainframe resources it used, based on measurements the system records.