Technical metrics are primarily concerned with the quality of the application's service, such as response times or throughput. They are the indicators that represent the application's responsiveness and may point out potential performance issues. The information can be used to create statistics about trends and application peaks.

This insight increases the likelihood of foreseeing potential outages and performance issues in a timely manner. It is the technical equivalent of business insights into the otherwise black box system. These metrics alone allow no sound conclusions about the root cause or constraining resources in the case of performance issues.

Lower-level technical information includes resource consumption, threading, pooling, transactions, or sessions. It's again important to mention that this information alone does not direct the engineers to potential bottlenecks.

As shown previously, it's necessary to inspect the overall situation with everything running on specific hardware. The operating system information provides the best source of information. In order to solve performance issues, the operating system, as well as application tools, are required to take this into account.

This doesn't mean that the technical information emitted by the application or the JVM runtime has no value at all. The application-specific metrics can assist in solving performance issues. It's important to keep in mind that these metrics alone will lead to potentially wrong assumptions about what the constraining resources are when a system needs to be performance-tuned.