Many Linux performance tools are underutilized -- a big problem with powerful tools is knowing what to do with them.
Are some Linux performance tools undervalued by system administrators?
On Linux, there are some obscure performance tools like pchar -- which characterizes bandwidth, latency and loss of links -- and less obscure but still undervalued ones like pidstat -- which monitors individual tasks that are managed by the Linux kernel.
The most undervalued tool is Linux perf_events, which does static and dynamic tracing. If you browse all the documentation you can find on perf_events, you're probably still only finding 1% of its capabilities.
Other Linux tracing tools are related to perf_events. A very promising one, ktap, is still in development, but may enter the mainline Linux tree.
In Systems Performance: Enterprise and the Cloud, I document new methodologies for approaching performance, which leads to needs and uses of those tools.