The AIX agent tracks CPU, memory, disk I/O, and network, and on AIX it adds the PowerVM micro-partitioning metrics that matter on shared-processor LPARs: Virtual CPU Usage, Physical CPU Steal, and CPU Entitlement. Every metric is collected automatically, normalized for comparison to Windows and Linux, and plotted against response-time degradation.
CPU, memory, and disk are captured in the Free edition on every monitored host. Network and TCP metrics, plus process queue length, are part of the Cloud full-metric set, alongside AWS CloudWatch and Dynatrace integration. The agent collects its full set automatically, with no metric-selection step to configure.
On a shared-processor LPAR, CPU percentage alone does not tell you whether the partition is actually getting the physical processor it needs. Physical CPU Steal shows cycles the hypervisor handed to other partitions in the shared pool. CPU Entitlement shows the processor capacity guaranteed to this LPAR. Virtual CPU Usage shows how the virtual processors are loaded. When response time degrades on an AIX tier, those three together separate "the application is slow" from "the pool starved this partition," which is a distinction you cannot make from utilization percentage on its own.
The Analytics Dashboard plots every collected metric on the same x-axis as response time and flags any resource whose curve correlates above 0.94 with the slowdown. On Cloud the AI report writes the bottleneck-source paragraph from those flags: which resource saturated, on which host, at which user level the response time went vertical.
Run a load test against your AIX hosts and watch the metrics line up.
The monitoring agent ships in every WPLoadTester 7 install and captures CPU, memory, and disk for free. The full network and TCP set is part of the Cloud edition. Download it to see your own LPAR metrics correlated against response-time degradation.
Comparing tiers? See the Free vs Cloud split.