The metrics required for effective capacity management (CM)
are many and various. Many sites start
with the basic component/resource metrics used primarily for event management
and extend them to provide a performance monitoring and reporting regime.
The starting point is to do it for normal production service, but it may need to be extended to identify the requirement for development, test and acceptance (or pilot) as well as production. And it may need to be developed to have less stringent values for periods of failover and disaster recovery.
If the metrics can be aggregated for processes into totals,
reflecting services, this can be used to provide more effective and meaningful
measures.
But to close the loop between the business view and the
machine room view, the key is to identify the business metrics, activity
drivers and business performance indicators.
These can then be reflected in the SLA with relevant (and measurable)
targets and constraints and then those mapped to the low level metrics
available from the operating system and any required application
instrumentation.
This can be mapped into a matrix for every service to be the
subject of full CM
The starting point is to do it for normal production service, but it may need to be extended to identify the requirement for development, test and acceptance (or pilot) as well as production. And it may need to be developed to have less stringent values for periods of failover and disaster recovery.
On Wednesday I’ll be tackling ‘How do I get there?’
Adam Grummitt
Distinguished Engineer
No comments:
Post a Comment