Indeed the metrics are different and can`t be compared. Usage
is a percentage of available machine memory, while Pressure
shows a potentional memory demand that is based on total allocated memory for running VMs, memory overhead, effects of memory Transparent Page Sharing and total available memory.We advise to set the alarm thresholds
to 110-120% for warning and 140-150% for error. In the next version we will make it our default values.