When you need it

As a linux/unix system administrator you may be asked to generate some graphs about the common system metrics (ie : cpu, memory, disks access, and so on...), and immediately you start with a number of tool to get the metrics, collect data and "purge" the data 'cause of indentation, output formatting, unwanted information or numeric format, etc...
You may use the SAR daemon tool to collect all the interesting metrics of a host and use the SADF command to format them in CSV or XML, and then ? use gnu graph ? or maybe excel ? or ksar ? Yes, you can, and automatically loose any ability to drilldown "interactively" on the data.

OR

you can import them in a quite more advanced tools such as QlikView or PowerBI, configuring ONE and only ONE interactive document with all the graph, metrics, filters, etc and only refresh the values from the collected ones only when needed. This is the road i want to follow : one interactive document which allow me to filter for time, disks access times, luns or volume, process and memory.
You could follow this road configuring all the system-collection-tools as you wish or you can give SARBI a try, at least you will start with a basic boilerplate to work on.

Guide

Samples

Use Cases