The role of ontologies in autonomic computing systems

IBM Systems Journal,  Sept, 2004  by L. Stojanovic,  J. Schneider,  A. Maedche,  S. Libischer,  R. Studer,  Th. Lumpp,  A. Abecker,  G. Breiter,  J. Dinger

• E-mail
• Print
• Link

The increasing complexity of information technology (IT) systems demands a correspondingly greater effort for systems management. Today, many systems management tasks such as system configuration, performance analysis, performance tuning, error handling, and availability management are often performed manually. This work can be time-consuming and error-prone. Moreover, it requires a growing number of highly skilled personnel, making IT systems costly. IBM's autonomic computing initiative, (1) which is a core element of IBM's e-business on demand * strategy, (2) addresses this problem by developing and providing powerful concepts for self-management, including new self-healing, self-protecting, self-optimizing, and self-configuring capabilities. The goal is to reduce the burden associated with the management and the operation of IT systems. Autonomic computing systems should simply work, repairing and tuning themselves as needed. This requires that such systems be able to protect themselves, to identify upcoming problems, and to make required reconfigurations dynamically in order to resolve problems.

As one important step toward a vision of autonomic computing systems, IBM has developed several correlation engines. (3,4) These include ABLE (Agent Building and Learning Environment), AMIT (Active Middleware Technology), eAutomation, TEC (Tivoli Enterprise Console *), Yemanja, and ZCE (Zurich Correlation Engine). Correlation engines are autonomic core components that perform continuous automated analysis of enterprise-wide, normalized, real-time event data based on user-defined configurable rules. These rules can be used to detect threats, complex attack patterns, or system failures, and to initiate a corresponding reaction.

Today IBM'S correlation engines are being used successfully in many application domains. Examples include the eAutomation correlation engine in System Automation for OS/390 * (5) and System Automation for Multiplatforms, (6) and the Zurich Correlation Engine for router fault isolation and for intelligent monitoring capabilities.

In this paper we propose the application of formal ontologies (7,8) as the conceptual backbone for correlation engines. In philosophy an ontology is a theory about the nature of existence and, in particular, about what types of things can exist; ontology as a discipline studies such theories. Artificial intelligence and Web researchers have adopted this term for their own purposes. For them an ontology describes a formal, shared conceptualization of a particular domain of interest. (7) Thus, ontologies provide a way of capturing a shared understanding of a domain that can be used both by humans and systems to aid in information exchange and integration. The use of ontologies has several advantages.

First, ontologies can facilitate interoperability between correlation engines by providing a shared understanding of the domain in question. In this way problems caused by structural and semantic heterogeneity of different models can be avoided. Structural heterogeneity results when different correlation engines store their data in different schemes. Semantic heterogeneity involves similar problems in the content and intended meaning of information. Ontologies provide an effective means for explicating implicit design decisions and underlying assumptions at system build time. This makes it easier to reason about the intended meaning of the information interchanged between two systems. Hence, interoperability is a key benefit of the application of ontologies, and many ontology-based approaches to information integration have been developed. (9)

Second, ontologies provide a formalization of shared Understanding which allows machine processability. Machine processability in turn forms the basis for the next generation of the World Wide Web, the so-called Semantic Web, (10,11) which is itself based on using ontologies for enhancing (annotating) content with formal semantics. This will enable autonomic agents to reason about Web content and to carry out more intelligent tasks on behalf of the user.

Finally, the explicit representation of the semantics of data through ontologies will enable correlation engines to provide a qualitatively new level of services, such as verification, justification, and gap analysis, as we discuss later in this paper. These engines will be able to weave together a large network of human knowledge and will complement this capability with machine processability. Various automated services will then aid users in achieving their goals by accessing and providing information in a machine-understandable form. It is important to note that ontologies not only define information, but also add expressiveness and reasoning capabilities. Ontology rules provide a way to define behavior in relation to a system model.

The focus of this paper is on the third, most ambitious benefit that can be achieved by using ontologies in autonomic computing, namely the providing of new levels of services. To illustrate this benefit we will use the eAutomation correlation engine technology and its resource relationship model.