Originally posted by: D.R.Saravanan

HMC version 810, released in June 2014, introduces a new cool feature for monitoring Performance of Managed Systems (Power Systems). This feature supports 3 functions:

1. Provides REST APIs to retrieve Compute & IO performance utilisation metrics in its raw form
2. Harvests performance utilisation metrics, rolls-up the harvested data for trending purposes. Provides REST APIs to retrieve the harvested data
3. Performance Monitor Graphical User Interface to visualise the harvested performance data

The HMC Performance Monitor function provides detailed set of performance utilisation metrics that is useful to customers to understand the systems & partitions health, behaviour and bottlenecks.

Following is the set of high level metrics exposed by the monitoring functions:

• Processor
  • Systems, Hypervisor(s) [ including per VIOS ] and per Partition utilization
  • Per Partition Dispatch Wait Time
• Memory
  • Systems, Hypervisor(s) [ including per VIOS ] and per Partition Assignment

Note: Only allocations to partitions will be reported & not the OS memory usage.

• Network IO
  • IO Hypervisor [ VIOSs ] Network Adapters utilization
  • Network Bridge utilisation with per Partition breakdown
  • Virtual Lan utilisation with per Partition breakdown
  • SRIOV adapter utilisation with per Virtual Function & Partition breakdown 

Note: Only Virtual Network utilisation metrics available at partition level

• Storage IO
  • IO Hypervisor [ VIOSs ] Storage Adapters utilization
  • vSCSI adapter utilisation with per Partition breakdown 
  • NPIV adapter [ virtual FC ] utilisation with per Partition breakdown

Note: Only virtual SCSI or vFC utilisation metrics available at partition level

The HMC Performance Monitor function is not intended to replace any higher level performance monitoring tools like Tivoli or BMC and also the solution is not intended to act as a data-ware house or a health centre. The HMCs storage capacity is limited, hence the amount of performance data stored in the HMC also be limited. In version 810 HMC can keep a max of 1 year of performance data in rolled-up form (1 sample per day) for each Managed System. 

Certainly the data can be retrieved from HMC, can be stored externally by users for any 3rd party or custom tool consumption. Users can retrieve the performance data from HMC using the provided REST APIs. HMC exports the data in JSON format. 

HMC 810 version doesn't provide a graphical user interface for the export functionality instead the REST APIs can be used for the same purpose. The details of the APIs can be found in the HMC Rest API documentation.

You can find the recorded demo of the HMC Performance Monitor GUI here or in the 'Files' section of this community.