Monitoring in Federated and Self-Manageable Clouds

Background

Regardless of the service model of any given Cloud, let it be IaaS, PaaS or SaaS, some common architectural components are shared in all. All Clouds can be seen as pools of resources, which serve requests for allocating those to their clients and making sure they receive a minimum Quality of Service. To realize that, each Cloud must have components dedicated to allocation and relocation of resources so as to meet the needs of their clients and ensure their smooth operation. Such components are decision making mechanisms which solve optimization problems over the set of available and required resources, costs and time requirements for placement and relocation. In addition to these, all Clouds contain components which are responsible for accounting the actions of their clients and billing them. Of course, depending on each individual Cloud’s feature list others may or may not be present. The Monitoring mechanism is required to exist in all Cloud platforms, since it offers a very basic functionality that is crucial for the Cloud operation.

The Monitoring mechanism's exact responsibilities vary in different Cloud platforms, but there is a common set of services provided. It's this module which monitors physical & virtual resources and produces events, aggregates those events, and notifies any other modules interested in them. It is very common to have the corresponding monitoring system perform additional tasks, like logging the events or extracting other, more high level events from the ones received, using some kind of intelligent mechanism.