Rule-based Knowledge Representation for Service Level Agreement

Automated management and monitoring of service contracts like Service Level Agreements (SLAs) or higherlevel policies is vital for efficient and reliable distributed service-oriented architectures (SOA) with high quality of service (QoS) levels. IT service provider need to manage, execute and maintain thousands of SLAs for different customers and different types of services, which needs new levels of flexibility and automation not available with the current technology. I propose a novel rule-based knowledge representation (KR) for SLA rules and a respective rule-based service level management (RBSLM) framework. My rule-based approach based on logic programming provides several advantages including automated rule chaining allowing for compact knowledge representation and high levels of automation as well as flexibility to adapt to rapidly changing business requirements. Therewith, I address an urgent need service-oriented businesses do have nowadays which is to dynamically change their business and contractual logic in order to adapt to rapidly changing business environments and to overcome the restricting nature of slow change cycles. 1. State of Art, Problem Definition and Research Question Service Level Agreements (SLAs) defining the performance criteria a provider promises to meet while delivering a service, are of growing commercial interest with a deep impact on the strategic and organisational processes. A well-defined and effective SLA correctly fulfils the expectations of all participants and defines the quality attributes and guarantees a service is required to process. It typically also sets out the remedial actions and any penalties that will take effect if performance falls below the promised service levels. During the monitoring and enforcement phase the agreed SLA rules will be used to detect violations to the promised service levels and to derive consequential activities in terms of actions, rights and obligations. They play a key role in metering, accounting and reporting and provide data for further refinement of SLAs and for optimizing service management on an operational, tactical and strategically level. In practice, IT service providers need to manage, execute and maintain thousands of SLAs for different customers and different types of services in the upcoming service-oriented computing landscape. Commercial service level management tools such as IBM Tivoli, HP OpenView, CA Unicenter, BMC Patrol, and Microsoft Application Center store selected QoS parameters such as availability or response time as parameters in the application code or database tiers. This approach is restricted to simple, static rules with only a limited set of parameters. Existing SLA specification languages such as WSLA, WSML or WSOL are pure mark-up serialization languages which need specialized procedural interpreters, have a restricted expressiveness to describe complex decision and contractual logic in terms of rules and provide no capabilities to declaratively implement new functionalities. As a result the upcoming service orientation based on services that are loosely coupled across heterogeneous, dynamic environments needs new ways of SLA representation with a high degree of agility and flexibility in order to efficiently manage, continuously monitor and enforce large amounts of complex and possibly distributed SLAs. I propose a novel declarative rule-based approach to SLA representation and management. Whereas, existing approaches to standardize SLAs are based on imperative procedural or simple propositional logic, I draw on logic programming (LP) and related knowledge representation (KR) concepts. LP allows for a compact representation of SLA rules and for automated rule chaining by resolution and variable unification, which alleviates the burden of having to implement extensive control flows as in imperative programming languages and allows for easy extensibility. However, further logical formalisms are needed for automated SLA management. I propose a novel, unique, highly expressive and efficient framework of adequate knowledge representation (KR) concepts. The underlying research question of this thesis is: Can advanced knowledge representation (KR) concepts using logic programming (LP) techniques and generic inference engines (rule engines) adapted to the SLA domain and will they enable an efficient, flexible and distributed management and enforcement of SLAs? Author Supervisor Dipl. Wirtsch.-Inf. Adrian Paschke Internet Based Information System (IBIS) Technische Universität München [email protected] Prof. Dr. Martin Bichler Internet Based Information System (IBIS) Technische Universität München [email protected] 2. Objectives and Contribution of the Work The proposed research is on the interplay of common software engineering providing well-established development methodologies and imperative program and database solutions, on the one hand and artificial intelligence and knowledge engineering following the declarative rule based programming paradigm, on the other. A third important research dimension is added due to the inherent business-driven character of the application domain, the indispensable compliance with legal regulations and the deep impact on all operational, tactical and strategic IT service management (ITSM) processes. The first field refers to the adoption of successful SE methodologies, principles and techniques such as agile test-driven development, object oriented imperative programming and SE principles such as data abstraction or modularization. For the reason of acceptance, (re)usability with respect to existing expertise and tools, and interoperability in a distributed environment it is vital for a practical SLA management tool to be based on established SE methodologies and reuse existing development tools and existing operational systems such as network and system management tools, relational databases / data ware houses or middle ware applications. The legal and business driven aspects of SLAs induce high requirements regarding reliability, verifiability and traceability of derived conclusions (e.g. derived penalties) and triggered reactions which must also count in the legal sense and fulfil legal regulations and compliance rules. Moreover, service-oriented businesses face rapidly changing environments which amount for frequently adapting the business and contract logic. Obviously, this amounts for an interaction between the technical layer and the business layer as well as between various roles reaching from technicians, business practitioners to domain experts and executives. The objective for the proposed research is to adapt the values, principles and practices in the addressed fields and bring together the different requirements within one common framework. I argue that my rule-based SLA approach using logic programming and inference engines as the “common denominator” is well suited to be the methodical and technical core for this goal. Clearly, traditional logic programming and standard inference engines need to be adequately extended with further expressive logical formalisms and KR techniques to represent sophisticated contractual logic and incorporate external data and systems into declarative rule descriptions. This thesis contributes with a declarative, compact and highly expressive KR proposal for formalizing SLA specifications, which has the following properties: • has a clear formal semantics, • is computationally feasible even for larger SLAs and data sets • reliable and traceable even in case of incomplete or contradicting knowledge • provides support for validation and verification of contractual specifications • is flexible in a way that allows to quickly alter the behaviour of the SLA system • supports programming of arbitrary functionalities and decision procedures • provides means for serialization and interchange based on a superimposed SLA mark-up language • enables reuse and integration of external data, systems and object-oriented code • is integrated in established development frameworks such as eclipse with Junit and Ant and service-oriented computing technologies such as Web Services and Semantic Web technologies The major advantages of my rule-based approach are: • it allows to continuously adapt SLAs to a rapidly changing business environment, and overcomes the restricting nature of slow change cycles • it lowers the cost incurred in the modification of contractual / business logic • it shortens development time of SLA specifications and safeguards the engineering process • rules are externalized and easily shared among multiple applications and contractual partners • changes can be made faster and with less risk • it provides are more human way to describe, manage and maintain formalized SLAs as compared to implicit procedural implementations • it provides highly reliable and traceable results derived by generic inference engines with logical semantics based on a solid mathematical basis • it allows to express and declaratively program highly sophisticated SLA logic and business management policies