Abstract: Methods and apparatus involve continuous management of workloads, including regular monitoring, profiling, tuning and fault analysis by way of instrumentation in the workloads themselves. Broadly, features contemplate collecting current state information from remote or local workloads and correlating it to predefined operational characteristics to see if such defines an acceptable operating state. If so, operation continues. If not, remediation action occurs. In a virtual environment with workloads performing under the scheduling control of a hypervisor, state information may also come from a hypervisor as well as any guest user and kernel spaces of an attendant operating system. Executable instructions in the form of probes gather this information from items of the stack available for control and deliver it to the management system. Other features contemplate supporting/auditing third party cloud computing services, validating service level agreements, and consulting independent software vendors.