Dynamic application placement

Dynamic application placement is based on load distribution, service policy, and available resources. You can run applications at various levels within a Intelligent Management environment, ranging from changing relative weights of applications to expanding the Intelligent Management cell. As changes become more drastic and their impacts become more extreme, applications take more time to run.

By dynamically placing resources, Intelligent Management allows you to utilize your hardware more efficiently. It is unlikely that all applications are in high demand at the same time, assuming a varied assortment of deployed applications. Intelligent Management exploits this situation by supporting resource allocation where needed, thereby increasing the utilization of hardware. The result is that an enterprise no longer requires enough hardware to satisfy each application maximum load simultaneously, which can translate to a significant reduction of requisite IT investments.

The dynamic application placement capability is contained by three autonomic managers: the application placement controller, dynamic workload manager (DWLM), and autonomic request flow manager (ARFM).

Each autonomic manager provides capabilities that work to achieve a common goal, which is to allocate the available capacity among the deployed applications and configured services classes. When it is necessary to modify the capacity that is allocated to a given application, the autonomic manager takes actions starting with the first or fastest level. As changing demand renders these lightweight adjustments ineffective, more drastic measures are required. For example, Intelligent Management autonomic managers might start by changing the dispatching priorities of requests that are associated with a service policy. Intelligent Management makes such adjustments every few seconds. However, such adjustments might be ineffective if the size of a given cluster is too small. Intelligent Management changes the sizes of dynamic clusters within tens of minutes.

The challenge of dynamically placing a given application includes increasing and decreasing its capacity, which can be accomplished at several levels. The decisions about the resource allocation changes are made using the autonomic managers, which monitor various performance metrics, analyze the monitored data, and then guide the planned actions. Intelligent Management supports various configurable levels of autonomicity. In the most autonomic mode, autonomic managers are granted the freedom to run the actions they plan. In the most manual mode, autonomic managers suggest actions, which an administrator must run manually.