The increasing scale, technology advances, and services of today’s networks have dramatically complicated their management in such a way that in the near future human administrators will not be able to achieve efficient management. To control this complexity, a promising approach aiming to create self-managed networks has been introduced. This approach, called autonomic computing, aims to design network equipment able to self-adapt their configuration and to self-optimize their performance depending on their situation, in order to fulfill high-level objectives defined by human operators.
The Autonomic NEtwork Management Architecture (ANEMA) implements several policy forms to achieve autonomic behaviors in the network.
To make networks autonomic, it is important to identify the high-level requirements of the administrators. These requirements can be specified as utility functions that illustrate, the practical and general way of representing them. At the next stage, it is necessary to map these utility functions to specific quality parameters and evaluate them based on information collected from the network. For that task, management architecture can employ some analytical optimization models. The specification of the constraints related to these functions and the way that they can be achieved can be guided by high level management strategies. These strategies permit defining a second form of autonomic network management rules, called goal policies.
These policies can be defined to guide the management process in the network elements to adopt the appropriate management behavior and to identify the constraints during the optimization of the utility functions. Some parts of goal policies are translated into abstract policies called behavioral policies, in order to help the autonomic entity to manage its own behavior in accordance with them. The last form of policies aims to describe the behavior the network equipment is to follow in order to reach the global desired state of the network. ANEMA aims at supporting all these mechanisms to achieve autonomic IP network management architecture.
The central problem in system and network management lies in the fact that the critical human intervention is time-consuming, expensive, and error-prone. The autonomic computing initiative (launched by IBM) seeks to address this problem by reducing the human role in the system management process. This initiative proposes a self-management approach that can be described as the capabilities of the autonomic systems and the underlying management processes to anticipate requirements and to resolve problems with minimum human assistance. To do so, the human role should be limited; the human should only have to describe the high-level requirements and to delegate control tasks that the system performs in order to achieve these requirements.
ANEMA defines an autonomic network management architecture that instruments a set of policy concepts to achieve autonomic behavior in network equipments while ensuring that the network behavior as a whole is achieving the high-level requirements from human administrators as well as the users as illustrated below;
As this figure shows, the ANEMA is organized into two layers: an objectives definition layer and an objective achievement layer.
Objectives definition layer: The main component in this layer is the ODP (Objectives Definition Point). It allows the administrators to introduce their high requirements and the experts to introduce high-level management guidelines. In the ODP, the high-level requirements are transformed into Network Utility Functions (NUF), whereas the management guidelines are transformed into abstract management strategies. The NUFs represent the policy rules that describe the network performance criteria from the human viewpoint. They are used to express the network functionalities in terms of optimization functions; whereas,the management strategies allow description of the high-level guidelines needed to map the NUF to a specific management architecture that can be implemented within the target network infrastructure.
Objectives achievement layer: this layer contains a set of GAPs. Each GAP is an entity that behaves in an autonomic manner while trying to achieve the target high-level requirements by considering the goal policies and the NUF optimization models. In fact, according to these informational elements, the GAP can take its own decisions to achieve the target requirements by means of its elementary monitoring, analyzing, planning, and executing (MAPE) capabilities. The GAP is also able to interact with its environment and communicate with other GAPs. In real networks, a GAP can be a router, switch, gateway, software, multimedia device, and so on.
