Operations Manager i 10.11 delivers a new component: The Performance Engine (PE). It is an add-on component which provides streaming of metrics (either custom or system metrics) in large scale environments. The Performance Engine is designed to collect metrics which are streamed in high frequency intervals. These metrics can be displayed using the OMi Performance Dashboard.
Real time metrics from the Operations Agent are stored for one hour (default). If a Vertica database is installed and configured, metrics can be stored for a much longer time-frame.
The Performance Engine requires an HPE Operations Agent version 12.01 and real time metric collection enabled. This can be done either manually per agent or using an agent profile during installation.
To manually enable real time metrics the according license needs to be set on the Agent:
Windows: %OvInstallDir%\bin\win64\oalicense -set -type PERMANENT "HP Ops OS Inst to Realtime Inst LTU"
Linux: /opt/OV/bin/oalicense -set -type PERMANENT "HP Ops OS Inst to Realtime InstLTU"
Afterwards, run ovc –restart
To enable the real time metrics for multiple agents an agent profile can be used. Create an agent profile and add the following lines to the profile: set eaagt.license:HP_Operations_OS_Inst_Adv_SW_LTU=PERMANENT set eaagt.license:HP_Ops_OS_Inst_to_Realtime_Inst_LTU=PERMANENT
Hint: For more details please consult the OMi Administration Guide: Configuring Metric Streaming Configuration Policies
Installation and Configuration
The Performance Engine is part of the OMi 10.11 installation bits. It supports RHEL6/7 as Operating System. It can be installed together with BVD (Business Value Dashboard) and DFP (Data Flow Probe).
Run the installer Software/HPEPerf_Engine_10.11.003_setup.bin
Run the configuration afterwards: /opt/HP/BSM/HPEPerfEngine/bin/admintools –configure This will ask you for a database password and integrate the PE with the Operations Agent.
Vertica Database (optional): In case metrics should be stored for a longer timeframe Vertica should be used as database. The required steps to use the Vertica Database are:
Install Vertica from the Operations Bridge Reporter media
Register Vertica Database in the Performance Engine: /opt/HP/BSM/HPEPerfEngine/bin/admintools –configuredb
Hint: For more details please see the "OMi Performance Engine Installation and Configuration Guide" which is part of the OMi 10.11 installation media.
As a next step the PE needs to be registered in OMi.
In OMi navigate to Admin -> Setup and Maintenance -> Infrastructure Settings
Open Applications: Performance Engine
Fill in the address of the Performance Engine and the Database password.
Hint: You can verify if the PE is connect with OMi with the following URL: https://<OMi server>/PE/rest/1.0/admin/endpoint
Configure Metric Streaming
Metric streaming is configured via policy. Starting from OMi 10.11, there is a new policy of type "Metric Streaming Configuration", which allows to select the metrics stored by the Operations Agent.
Hint: The policy can be deployed to an individual Operations Agent or to multiple Operations Agents.
Navigate to: Admin -> Monitoring -> Policy Templates Create a new policy of type "Metric Streaming Configuration":
When you click on "Load Metric Definition" you will see the Agent metrics which are available for streaming configuration:
I am using "GBL_CPU_TOTAL_UTIL" metrics.
"Save and Close" the policy and deploy it to the Operations Agent:
This is all you need to do. Once the policy is installed on the Operations Agent the metric streaming will immediate start.
Hint: You can verify if the policy is installed by executing ovpolicy -list on the Operations Agent
Now it is time to see the Performance Engine in action. The metrics can be visualized in an appealing way using the OMi Performance Perspective.
In OMi navigate to Workspace -> Performance Perspective. Click on the node for which you have enabled the metric streaming and create a new dashboard.
As data source select the "Operations Bridge Store", this is where the Performance Engine metrics are stored. Also select the "Class Name", "Metric Name" and "Instance Name" from the list and click "Back to dashboard".
From the "Settings" you can enable "Raw data points":
In the "Data Range Panel" select the timeframe, e.g. the last 15mins.
Congratulations: You have created your first metric streaming dashboard for the CPU metric:
Hint: The default publish interval for Operations Agent is 10 seconds. This can be changed to a minimum of 5 seconds. Run the following command directly on the Operations Agent: ovconfchg –ns hpsensor –set PUBLISH_INTERVAL 5
The Performance Engine is a great add-on component which allows you to store high frequent metrics for a longer time frame. You can investigate system metrics based on real time data even for past events.
In addition the Performance Dashboard helps to visualize, investigate, and review the performance of specific system infrastructure metrics as well as custom application metrics in real-time.
Try it out now:
Download Operations Manager i 10.11 and get started with the Performance Engine.