analytics - analytics BUI
The Solaris Analytics BUI provides the ability to visualize statistics and diagnose common performance problems for a single Oracle Solaris system. This BUI can be accessed through the Oracle Solaris Dashboard, through the Applications menu, or directly through the following URL:
The Analytics BUI provides a pre-defined set of sheets that provide different views of the system. Each sheet is comprised of a set of sections, which in turn contain groups of visualizations. You may assemble your own sheets with a combination of custom visualizations and visualizations defined by the system. Visualizations allow multiple visualization types for statistics and events provided by the Oracle Solaris Statistics Store, sstore. For more information, see the sstore(7) man page.
Each sheet provides a live view of the system by default, and visualizations allow interaction to show different time horizons or views on the data.
Many of the sheets pre-defined by the Analytics application reflect utilization, saturation, and errors of key system resources.
Utilization is expressed as the currently achieved percentage of theoretical maximum throughput. Saturation is the amount of work queued, that is, not actively serviced. It is sometimes expressed as a multiple of the maximum utility (capacity). Errors are occurrences per second, and these should usually be low or zero.
More general sheets and visualizations show these key metrics across all available resources for that measurement. This provides an overall appreciation of the state of that resource. For example, if the average CPU utilization is at 80%, we can further explore why CPU utilization is so high. Any saturation (work in excess of capacity) is an indication of overload, and any errors are typically cause for immediate investigation.
In addition to the average metrics, individual resource utilization is shown for many resources. This is because resources are commonly allocated directly to workloads (such as CPUs to zones or NICs on distinct networks). Analytics also provides partitioned views to quickly identify whether individual resources are seeing excessive utilization, saturation or errors.
Events are also often shown inline with performance graphs. These might include administrative changes to relevant resources such as configuring a new datalink, or FMA faults such as a CPU being proactively retired due to persistent problems.
The default view of the Analytics application is a list of sheets, which collects performance information for specific resources, or for specific diagnosis tasks. From this view, a sheet can be opened by clicking on the graph icon, or the sheet's information can be viewed by clicking elsewhere within the sheet area.
Filtering, sorting, and grouping allow exploration of sheets.
You can add new sheets, and also delete sheets that you own. Sheets delivered by the system cannot be deleted.
A sheet is comprised of a set of sections, containing groups of visualizations. You can only modify sheets that you own. The properties of the entire sheet can be viewed by selecting the Properties button.
All visualizations on the sheet can be paused to allow more detailed exploration of the data shown without the distraction of updates.
You can add a visualization, remove a visualization, copy a visualization from another sheet, or modify a visualization that you own. Favorite a visualization to make it easier to add that visualization to a new sheet.
The properties of the visualization, including which statistics (ssid) are on the visualization, are viewed and modified from the "settings" icon on each visualization. On modifiable visualizations, adding a Statistic in the Properties dialog displays a statistic browser. For more information, see the ssid(7)) man page.
Clicking on an existing statistic enables you to view and edit the SSID and make formatting choices for that statistic in the graph.
Some statistics can be partitioned. If partitions are available for a statistic, a partition dropdown will appear to allow selection. Partitions provide a breakdown of statistics by their component resources. For example, CPU usage can be partitioned by mode, or physical and virtual groupings (core, chip, pg, lgroup).
Charts and sheets can be delivered in IPS packages, in addition to being defined through the BUI. See the analytics(5) manual page for more details on the file format and delivery locations.