
01df0e23e26ddbf77be16769f58304ca.ppt
- Количество слайдов: 20
Definition of Metric Dependencies for Monitoring the Impact of Quality of Services on Quality of Processes Christian Mayerl, Kai Moritz Hüner, Jens-Uwe Gaspar, Christof Momm, Sebastian Abeck Cooperation & Management, Institute of Telematics, Faculty of Computer Science, Universität Karlsruhe (TH) 76128 Karlsruhe, Germany {mayerl | huener | gaspar | momm | abeck}@cm-tm. uka. de
INTRODUCTION – Overview n Introduction n Related Work n Metric Dependency Model Elements for modeling metric dependencies o Formalism for calculating metric dependencies o n Example – Performance Metric Dependencies Dependency model of response time o Application of response time metrics o n Monitoring Architecture Components processing metric definitions o Implementation experiences and evaluation o n Conclusion and Outlook 21 st May, 2007 IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 2
INTRODUCTION – Scenario and Questions Process Logic Business Process Activity To. RProcess. Collect. Student. Information Service related metric dependency functional dependency Service Operation Student. DBService. get. Complete. Set Application Logic related Metric. Service metric dependency functional dependency Component Function Student. DBAccess. get. Complete. Set Metric. Process related Metric. Component n Which elements needed to define dependencies between metrics o related to processes, invoked services and components implementing services? n How to define metric dependencies to allow o 21 st May, 2007 mapping of metric dependency definitions to an monitoring architecture? IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 3
RELATED WORK – Overview n Basics of quality definitions o UML Profile for Modeling Quality of Service n o basic terminology and structure, no calculable quality dependencies Quality of Web Services n raw and aggregated metrics, no dependencies between services and processes n Metric dependencies o Business-driven IT Management (BDIM) n o optimize SLOs from business perspective, not monitoring dependencies at runtime Dependencies between service and process metrics n e-business processes based on Web services, no management function to monitor Quality of Service (Qo. S) and to calculate the impact on Quality of Process (Qo. P) n Management architectures and applications o Common Information Model (CIM) – Metrics Model n 21 st May, 2007 metrics related to functional units, no definition of calculable metric dependencies IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 4
RELATED WORK – Common Information Model (CIM) Metrics Model CIM_Managed. Element * CIM_Metric. Def. For. ME CIM_Managed. System. Element * * Validity : uint 16 * CIM_Logical. Element CIM_Unit. Of. Work. Definition ID : string {key} Context : string Trace. Level. Type. ID : string 1. . * CIM_Uo. WMetric. Definition * CIM_Base. Metric. Definition CIM_Uo. WMetric Value : string Modeling of runtime management information o 21 st May, 2007 CIM_Uo. WMetric * * CIM_Unit. Of. Work ID : sint 64 {key} Uo. WDef. ID : string {key, propagated} User. Name : string * Start. Time : date. Time Elapsed. Time : datetime Status : uint 16 {enum} Mutual. Context. Id : string Trace. Level : uint 32 0. . 1 CIM_Sub. Uo. W Id : string {key} Name : string Date. Type : uint 16 {enum} Calculable : uint 16 {enum} 1 Units : string Breakdown. Dimensions : string[] CIM_Metric. Instance Is. Continous : boolean Change. Type : uint 16 {enum} Time. Scope : uint 16 {enum} Gathering. Type : uint 16 {enum} n 1 * CIM_Logical. Element. Performs. Uo. W * CIM_Metric. Definition * CIM_Started. Uo. W Id : string {key} Metric. Definition. Id : string {required} Measured. Element. Name : string Time. Stamp : datetime Duration : datetime Metric. Value : string Breakdown. Dimension : string Brek. Down. Value : string Volatile : boolean CIM_Logical. Element. Unit. Of. Work. Def CIM_Sub. Uo. WDef CIM_Base. Metric. Value … CIM_Metric. For. ME * … Metric dependencies associated with relation CIM_Sub. Uo. W between units of work IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 5
METRIC DEPENDENCY MODEL – Assumptions and Requirements n Assumptions o o o focus on one process activity and one invoked service operation process implemented with Business Process Execution Language (BPEL) services described with Web Service Description Language (WSDL) services implemented by component-based applications functional dependencies between services and processes are well-known n Requirements provider monitoring Quality of Services (Qo. S) should be able to estimate impacts on Quality of Processes (Qo. P) o formal model of metric dependencies needed to be calculated o mapping on existing standard management information models (e. g. Common Information Model, CIM) o 21 st May, 2007 IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 6
METRIC DEPENDENCY MODEL – Metrics and Metric Dependencies Process Layer 1. . * * * Partner. Link Activity 1 * 1 Metric. Value Structured. Activity 1. . * Basic. Activity 1. . 2 1 Role Invoke. Activity Service Layer * depends on 0. . 1 1 * Interface. Operation 1 * * 1. . * 0. . 1 Metric. Category * 21 st May, 2007 depends on Application Layer * Component. Function * Metric. Definition 1 Unit. Of. Work. Definition 1 * * Service * Unit. Of. Work 1. . * * Metric. Characteristic * 1. . * Dependency. Function Metric. Dimension * 1 IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 7
METRIC DEPENDENCY MODEL – Formalizing Metric Dependencies Functional Entity Quality Value Metric Value 21 st May, 2007 IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 8
EXAMPLE – PERFORMANCE METRIC DEPENDENCIES Introduced Example Scenario Process Logic Business Process Activity To. RProcess. Collect. Student. Information functional dependency Service Operation Student. DBService. get. Complete. Set Application Logic functional dependency Component Function Student. DBAccess. get. Complete. Set n Process activity invokes a service operation n Service implemented by application component functions 21 st May, 2007 IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 9
EXAMPLE – PERFORMANCE METRIC DEPENDENCIES Metric Dependency Pattern for Performance request received processing started response sent response received Application Layer Service Layer Process Layer request sent n Metric dependency patterns based on existing approaches n Pattern applied to specific SOA by assigning concrete architectural elements to each metric 21 st May, 2007 IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 10
EXAMPLE – PERFORMANCE METRIC DEPENDENCIES Model for Student. DBService and To. RProcess : Metric. Category : Process name = “To. RProcess” : Structured. Activity : Invoke. Activity name = “Lookup. Student” name = “Performance” : Unit. Of. Work. Definition name = “Lookup. Student” : Metric. Characteristic name = “main” name = “Duration. Time” : Unit. Of. Work. Definition name = “use. Network” : Service name = “IStudent. DBService” name = “Transmission. Time” : Dependency. Function name = “Student. DBService” : Interface : Metric. Characteristic name = “Calc. Duration. Time” : Interface. Operation name = “is. Student” : Unit. Of. Work. Definition name = “is. Student. Op” : Metric. Characteristic name = “Response. Time” : Dependency. Function : Component name = “Calc. Response. Time” name = “Student. DBAccess” : Component. Function name = “is. Student” : Unit. Of. Work. Definition name = “is. Student. F” : Component. Function name = “get. Complete. Set” 21 st May, 2007 IEEE/IFIP International Workshop on Business-driven IT Management : Metric. Dimension name = “Waiting. Time” : Metric. Dimension name = “Processing. Time” Dr. Christian Mayerl 11
EXAMPLE – PERFORMANCE METRIC DEPENDENCIES Simple Metric Dependency Function n Assumptions n Dependency functions 21 st May, 2007 IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 12
MONITORING ARCHITECTURE – Component Overview Deployed on BPEL Engine «component» Management Port «component» Process Manager BPEL Process Get Metrics Establish Agreement «component» «delegate» «component» Agreement Repository Provide Report Notify Impact Agreement Monitor Provide Agreement «delegate» «component» Metric Comparator Report Generator «delegate» Management Port «component» Application Server Provide Service (WSDL) «component» Web service Implementation 21 st May, 2007 Provide Metrics Metric Aggregator «delegate» «deployed on» Implemented as CIM Provider Management Port Implemented as CIM Server «component» Metric Collector IEEE/IFIP International Workshop on Business-driven IT Management Store & Provide Data Storage Dr. Christian Mayerl 13
MONITORING ARCHITECTURE – Component Details n Agreement. Monitor o Metric. Aggregator n o Metric. Comparator n o calculating metric values based on functional and metric dependencies comparing SLOs with calculated metric values Metric. Collector n collection of metric values and transformation to management information standards (e. g. CIM) n Others o Storage n n o Instrumentation n n 21 st May, 2007 Archiving of measured values Associating values with functional and metric entities manageability capabilities of functional elements provision of sufficient management information IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 14
IMPLEMENTATION EXPERIENCES – Demonstrator «device» : Alfi. VM 02 «device» : Alfi. VM 03 «execution. Environment» : Oracle. BPELengine To. RProcess. jar «execution. Environment» : Axis 2 web. App «artifact» Agreement. Monitor. Service. aar «artifact» Agreement. Monitor. jar «device» : My. SQLServer courseunitdata examinationdb … «artifact» Agreement. Service. aar «artifact» Agreement. Repository. jar «artifact» Report. Generator. jar «artifact» Metric. Comparator. jar «artifact» Metric. Aggregator. jar «device» : Alfi. VM 01 «execution. Environment» : Axis 2 web. App Course. Unit. DBService. aar Examination. DBService. aar Organization. DBService. aar Student. DBService. aar «artifact» Dimension. Value. Service. aar «artifact» CIMOMwrapper. jar «artifact» Metric. Dependency. Mgmt. Service. aar «artifact» CIMOMwrapper. jar «execution. Environment» : CIMMOM CIM_Unit. Of. Work. mof Qo. S_ … … n 10. 000 requests for collecting and aggregating metric values 21 st May, 2007 IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 15
IMPLEMENTATION EXPERIENCES – Evaluation of Metric. Collector and Metric. Aggregator n Useful medians 141 milliseconds for Metric. Collector o 437 milliseconds for Metric. Aggregator o 21 st May, 2007 IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 16
milliseconds IMPLEMENTATION EXPERIENCES – Evaluation of Metric Dependencies (Component, Service, Process) 21 st May, 2007 IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 17
milliseconds IMPLEMENTATION EXPERIENCES – EVALUATION Sorted Measurement Data 21 st May, 2007 IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 18
Service Component IMPLEMENTATION EXPERIENCES – EVALUATION Correlation of Measurement Data Process Component Process Service 21 st May, 2007 IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 19
CONCLUSION AND OUTLOOK n Conclusion o Modeling metric dependencies between elements of a SOA n n o Formalism to calculate metric dependencies Example – response time dependencies Monitoring quality impacts at run time n n Development of a monitoring architecture Evaluation of monitoring metric dependencies n Outlook o Definition of a UML profile to model metric dependencies n n o Extensions of the management instrumentations n n 21 st May, 2007 Support for development of manageable SOA elements Integrated development of SOA and manageability capabilities Development of adequate management interfaces Provision of management information for calculating metric dependencies IEEE/IFIP International Workshop on Business-driven IT Management Dr. Christian Mayerl 20