Performance metrics

The HP Operations agent embedded performance component (EPC; also known as coda) collects performance counter and instance data. You can use these metrics in defining event/action thresholds that generate alarms in real time based on availability, response time, and throughput measurements.

The following types of metrics are collected:

Basic (golden) metrics
These are approximately 30 metrics that are collected for all supported platforms. They can be used to answer most of your questions about a system's global configuration, CPU, disk, swap, and memory usage and have been chosen to offer the best information for the widest number of platforms.
Additional metrics
The data collection component also provides you with additional performance metrics on each of the supported platforms. Although these metrics vary by platform, they are available on most platforms and are generally useful for drill down and diagnosis on a particular system. On HP-UX, for example, there are approximately 120 of these additional metrics.

The collection interval is five minutes. All metrics, including golden metrics and the additional metrics, are collected. The data is kept in the data store for up to five weeks, at which time a week's worth of data is rolled out.

You can view a of list of available metrics in the HP Performance Agent Dictionary of Operating System Performance Metrics which is available at http://ovweb.external.hp.com/lpe/doc_serv/. (Select the product Performance Agent, the required version, OS, and language.)

Related Topics: