Monitoring of Business Processes in the EGI

Monitoring of Business Processes in the EGI Radoslava Hristova Faculty of Mathematics and Informatics, University of Sofia St. Kliment Ohridski, 5 James Baucher, 1164 Sofia, Bulgaria radoslava@fmi.uni-sofia.bg Abstract. The European Grid Infrastructure (EGI) uses partially serviceoriented grid middleware for grid computing (g-lite). In the context of SOAbased business process management the definition, monitoring and optimization of a business process in the infrastructure are still not supported. In this article we present an approach for business process monitoring in EGI, based on the service-oriented BPM platform. Keywords: grid, monitoring, business processes Introduction The business process is a set of logically-related tasks performed to achieve a defined business outcome [1]. Business prosess management (BPM), supports business processes using methods, techniques, and software to design, enact, control, and analyze operational processes involving humans, organizations, applications, documents and other sources of information [2]. The service-oriented architecture (SOA) [3] is an architectural style for developing systems and applications. Basic characteristics of the model are well-defined logical entities called services, which can be independently used. Applaying service-oriented style for BPM will improve the design and management of the business processes and will optimize their usage. Combining business process management with the serviceorientated architecture will provide flexibility and optimization to the developed business processes and reuse of existing assets. Evenmore the service-oriented approach is the preffered approach for building cloud systems the next grid generation. In [4] the authors proposed such service-oriented generic resource framework for cloud systems, which represents datacenter resources in a uniform way, allowing generic administration without knowledge of the underlying resource access protocol. The SOA lifecycle (Figure 1) consists of four phases: model, assemble, deploy and manage. We are describing them with respect to business process monitoring. During modeling phase are gathered requirements. In this phase the business processes are designed. If the business process will be monitored, key performance indicators (KPIs) are defined. Evenmore, simulations of the business process with the defined key indicators can be done. During assembe phase, new assets are developed or existing assets are used. The business processes are composed and the key

2 Radoslava Hristova performance indicators are implemted. During the deploy phase the developed business processes and key indicators are deployed into the process server and monitoring server. Thus, the people, process and information are integrated. In the last phase of the cycle the developed business processes, services and applications are managed. The defined business metrics are monitored. If during the SOA lifecycle, some improvements can be done, the business process is optimized and processed through the lifecycle again. Fig. 1. Service-oriented architecture lifecycle The European Grid Infrastructure [5] (EGI) uses partially service-oriented grid middleware for grid computing (g-lite) [6]. In the context of SOA-based business process management the definition, monitoring and optimization of a business process in the infrastructure are still not supported. In [7] the author discusses the important aspects of service-orientated grids and underlines the lack of widely accepted mechanisms for business process orchestration, mediation and monitoring in it. G- Lite is not an exception. The goal of current investigation is to present an approach of business process monitoring for the EGI, based on service-oriented BPM platform for the EGI. Tools in the EGI and their service-orientation In [8] we present some of the tools, which are available for g-lite and can be used for building and executing service compositions in the EGI. All of the presented tools use their one mechanism for service composition, independently from g-lite. We will focus on three of these tools, which fulfil the requirements for service-orientation: Triana [9], Taverna [10] and Kepler [11]. Our choise is influenced and from [12], where the same tools are described as grid tools for monitoring and control of the workflow execution. We are describing the tools with respect to the features they provide for business process monitoring. The software Triana is a distributed environment, which provides functionality for building and executing business processes (workflows). The software is adapted for access to the grid resources, including the g-lite middleware. From architectural point of view, the environment consists of three major components: user interface, Triana

Error! Use the Home tab to apply title to the text that you want to appear here. 3 service and Triana control service. According [13] Triana could be used as a visual environment for monitoring the workflow of grid services. The user can get the information about the time for which the workflow is executed and the information about the status of the job (done, running, etc.). Taverna software provides environment for business process design and execution in grid. From architectural point of view, the software consists of two major modules: Taverna Workbench and Taverna Engine. The Taverna Workbench is a graphical editor for business process modeling, which provides functionality for monitoring of the designed process. On (Figure 2) is shown example of progress report for workflow executed into the EGI. The monitoring gives information for the status and the average time for execution of the workflow. Fig. 2. Taverna monitoring tool Kepler is a tool for design of business processes (workflows). The tool provides a graphical interface for business process modeling. The business processes can be described by using five basic components: directors, actors, parameters, link and ports. The director controls the execution of the process. The actor provides realized functionalies and follows the director s instructions. Every process has exactly one director. In [14] authors discussed how grid workflow designed in Kepler can be monitored. The monitoring process gives information for the jobs that are executed on the grid. The users should be able to view the status of the workflow (submitted, active, done, etc.), the tasks currently executing, and other information. The three envirionments Triana, Taverna and Kepler support service-orientation and can be used for workflow execution in the EGI. Unfortunattly, the monitoring tools wich these envirionments provide, do not allow user to customize, define or choose his own indicators for monitoring. That is not the case for SOA-based BPM systems. In the next section we present such system and example solution for this problem. SOA-based BPM Platform for EGI In [15] we present a framework for service-composition in the EGI. The software implementation of the framework is based on a SOA-based BPM Platform, presented on (Figure 3). On the first layer of the model are the legacy EGI services and applications. The layer covers already built it grid infrastructure, i.e., all the EGI services and applications are available and they can not be changed. On the second layer are all of the developed web services for access to the EGI services and application. The web services are important part form the framework.

4 Radoslava Hristova They participate into compositions in the higher layers. Additional layer is implemented in order to do the model applayable for the EGI. The module provides adapters for access to the EGI services and applications and exposed them as web services. The current implementation of the module provides web services for access to g-lite and web services for access to ROOT application. The developed web services are composable. Fig. 3. Architecture of a SOA-based BPM platform for the EGI On the third layer are the registry services, which provide features for publishing and discovery of the developed web services from the second layer. The access to the service registry can be done through the development module and through the runtime module. In [16] the authors present comparison of grid resource discovery approaches. Based on the functional requirements that they defined, we can conclude that, the UDDI approach is the most appropriate approach for grid resource discovery with respect to the defined criterions. The SOA-based BPM system also relies on the UDDI. Our implementation supports the UDDI solution for service registry. On the forth layer is the runtime envirionment module of the framework, which includes the enterprise service bus, business-process management service, business process monitoring services, human tasks services and business rules management services.

Error! Use the Home tab to apply title to the text that you want to appear here. 5 The fifth layer of the framework includes development module. It covers instrumental tools for design and deployment of the business process into the runtime envirionment module. The business processes can be designed, executed, managed and monitored in the forth and the fifth layers from the framework. In this article we are focussing on the monitoring services from the runtime envirionment module of the framework. For the implementation of the service we used IBM WebSphere Business Monitor tool [17]. IBM WebSphere Business Monitor is business activity monitoring software that provides monitoring of events in real time by providing visual display of business process status, together with alerts and notifications. The users can create KPIs without requiring a new IT development and deployment cycle. They can expand KPI calculations, including calculations based on the relationship with other KPIs and also can receive automated alerts for warnings. The monitoring of the business process starts from the design module, where the business process is designed. The grid user can set KPIs during modeling phase. After that the business process is translated to the assemble phase, where some business assets are reused or implemented. For the defined KPIs in the modeling phase KPIs in the assembly phase are generated. Examples of KPIs in the deployment module are shown on (Figure 4). The grid user can choose either to monitor the start time of the process, the average working duration or the end time. With this solution more monitoring information can be collected and more information can be recieved. Fig. 4. Monitoring templeates for KPIs for the business processes

6 Radoslava Hristova The developed business process is deployed into the runtime envirionment, where the monitoring service and business process management service executes the deployed business process. The result from the excution and monitoring is displayed into the business space portal which is part form standart distribution of the IBM WebSphere Business Monitor tool. Example output is shown oh (Figure 5) Fig. 5. Business space business process monitoring The gauges are one of the ways to visualize information by representing KPI values on a gauge. A dial is used to represent the position of the KPI value relative to the range and target of the KPI. A needle indicates the current value of the KPI. The gauge view focuses on representing KPIs that belong to aggregated business measures in a monitor model. Each gauge represents the value of a single KPI. Conclusions In this article we present an approach for business process monitoring in the EGI, based on the service-oriented BPM platform. With this solution more monitoring information can be collected and the grid user can receive more information, which can be used for improvements and optimisations of the business processes. Acknowledgement This paper is supported by Sofia University St. Kliment Ohridski SRF under Contract 134/2012. References 1. Davenport, T., Short, J. "The New Industrial Engineering: Information Technology and Business Process Redesign, 1990 2. van der Aalst, M., at all., Weske, M. "Business Process Management: A Survey, 2003, http://bpt.hpi.uni-potsdam.de/pub/public/paperarchive/bpm2003.pdf 3. Keen M., at all., Patterns: SOA Foundation - Business Process Management Scenario, International Technical Support Organization, 2006, http://www.redbooks.ibm.com/redbooks/pdfs/sg247234.pdf

Error! Use the Home tab to apply title to the text that you want to appear here. 7 4. Zhelev R. and V. Georgiev. A Generic Resource Framework for Cloud Systems, Proceedings of The 4th International Conference on Distributed Computing and Gridtechnologies in Science and Education, June 28 - July 3, 2010 Dubna, Russia., pp. 268 278. 5. European Grid Infrastructure, http://www.egi.eu/ 6. g-lite, http://glite.cern.ch/ 7. Dimitrov, V. T., Development of applications with service-oriented architecture for grid, ACM New York, 2008, Proceedings of the 9th International Conference on Computer Systems and Technologies (CompSysTech '08), Article No.14 8. Goranova R. D., Service composition tools in g-lite, Conference Proceedings of the Fifth International Conference ISGT, 2011, pp. 228-235 9. Triana 4 User Manual, http://www.trianacode.org/docs/userguide/userguide.pdf 10. Taverna 2 Architecture, http://www.taverna.org.uk/developers/taverna-2- x/architecture/ 11. Getting Started with Kepler, https://code.kepler-project.org/code/keplerdocs/trunk/outreach/documentation/shipping/2.3/getting-started-guide.pdf 12. L. Kirchev, V. Georgiev and K. Boyanov, Workflow Management for a General Purpose Grid Platform of Commodity Computers, in Proceedings of the International Workshop on Network and GRID Infrastructures, Sofia, Bulgaria, 27-28 September, 2007. pp. 42 50. 13. Wang, I., Taylor, I., at all. Triana as a Graphical Web Services Composition Toolkit, Cardiff University, 2003 14. Nandita Mandal, N., Deelman, E., at all. Integrating Existing Scientific Workflow Systems: The Kepler/Pegasus Example, USC Information Sciences Institute, 2007, http://pegasus.isi.edu/publications/kepler-works07.pdf 15. Goranova, R. D., Framework for service composition in g-lite, American Institute of Physics, Conference Proceedings Volume 1404, 2011, pp. 218-224, ISBN 978-0- 7354-0976-7, ISSN 0094-243X. 16. Pashov G., K. Kaloyanova, Comparison of Grid Resource Discovery Approaches, Third International Conference on Information Systems & Grid Technologies, 28-29 May 2009, Sofia, Bulgaria, pp 138-147 17. WebSphere Business Monitor V6.0.2 - Features and usage scenarios, IBM Corporation, 2007