Question 1

What does load average actually measure?

Accepted Answer

On Linux it is the average number of processes that are either running or waiting to run (and, unlike some systems, also those in uninterruptible I/O wait). The key is to divide by your core count: a load of 4 is fully busy on a 4-core machine but heavily overloaded on a single core. Per-core load near 1.0 means fully utilized; above 1.0 means work is queuing and waiting.

Question 2

Is a high load average always bad?

Accepted Answer

Not necessarily. Per-core load up to about 1.0 means the CPUs are well used with little waiting - efficient, not broken. Sustained per-core load above 1.0 means processes are waiting for CPU, so latency rises. On Linux a high load can also come from processes stuck in disk I/O wait rather than CPU, so pair this with a look at CPU versus I/O before assuming you need more cores.

Question 3

What do the three numbers mean?

Accepted Answer

They are the load averaged over the last 1, 5, and 15 minutes. Comparing them shows the trend: if the 1-minute figure is well above the 15-minute, load is rising and a spike is building; if it is below, a spike is passing. A steady set of three similar numbers means a stable workload.

Deploy, monitor, and recover.

Control production changes.

Tools for application teams.

Company and product information.

Load average interpreter.

A number that only means something per core

Free tools for the backend you run.