Descripción del proyecto
THE SCIENTIFIC WORKFLOW COMMUNITY HAS RECENTLY IDENTIFIED A SET OF MAIN TOPICS THAT SHOULD BE ADDRESSED IN ORDER TO FULLY EXPLOIT THE CAPABILITIES OF SUCH TECHNOLOGY TO DEAL WITH THE COMPLEX COMPUTATIONS REQUIRED BY CURRENT SCIENTIFIC APPLICATIONS, THE MAIN OBJECTIVE OF THE PROJECT IS TO EXTEND AND ADAPT THE MODELS AND TECHNIQUES DEVELOPED BY THE RESEARCH TEAM IN THE LAST YEARS FOR THE DOMAIN OF (SEMANTIC) BUSINESS PROCESSES (DENEB FRAMEWORK) AND SCIENTIFIC WORKFLOWS (DVEGA FRAMEWORK), BASED ON THE PARADIGM OF THE NETS-WITHIN-NETS FORMALISM (BELONGING TO THE FAMILY OF OBJECT NET FORMALISMS), IN ORDER TO DEAL WITH SOME OF THE MOST IMPORTANT STATED CHALLENGES, THE PROJECT WILL FIRST STUDY IN DEPTH THE USE OF DENEB-DVEGA TO DEAL WITH SCIENTIFIC WORKFLOW ADAPTION MECHANISMS, TO DO THAT, A NOTATION FOR SCIENTIFIC WORKFLOWS AND A METHODOLOGY FOR THEIR DESIGN WILL BE PROPOSED, PROVIDING WITH A CATALOGUE OF PARALLEL PROGRAMMING PATTERNS, FORMALLY SPECIFIED BY MEANS OF OBJECT PETRI NETS, AND USING LINDA AS A COORDINATION MECHANISM ORTHOGONAL TO THE COMPUTING UNDERLYING MODEL, THE SET OF PROPOSED PATTERNS WILL INCLUDE ADAPTATIONS OF PREVIOUSLY DEVELOPED PROPAGATION AND EXCEPTION HANDLING COMPONENTS ALREADY PRESENT IN DVEGA, AMONG OTHERS, THE PROJECT WILL RECONSIDER THE ATOMICITY CONCEPT COMING FROM THE CONTEXT OF DATA-FLOW ORIENTED MODELS AS WELL AS THE ADDITION OF NEW PARALLEL PATTERNS AS STREAMING, FOR INSTANCE, THE ROLE OF DATA PROVENANCE AND CHECKPOINTING TECHNIQUES WILL BE EXPLORED IN ORDER TO ROLLBACK TO A PREVIOUS CONSISTENT STATE, THESE TECHNIQUES WILL ALSO BE USED TO ALLOW THE REUSE OF ALREADY DONE EXECUTIONS, A SECOND PART OF THE PROJECT WILL FOCUS ON DEFINING A FRAMEWORK TO ALLOW THE DEPLOYMENT OF THE DEFINED WORKFLOWS IN THE IMPLEMENTATION INFRASTRUCTURE, PARALLEL SOFTWARE THAT FULLY EXPLOITS THE HARDWARE IS RARE AND MOSTLY LIMITED TO THE SPECIALIZED AREA OF SUPERCOMPUTING, EFFICIENT EXECUTION OF SCIENTIFIC WORKFLOWS DEPENDS ON HOW WELL THE SOLUTION MAPS ON THE INFRASTRUCTURE, KEYWORDS SUCH AS INTEROPERABILITY, SCALABILITY OR AUTONOMOUS RECONFIGURATION WILL BE CONSIDERED, THE WAY OF ACHIEVING THIS WILL BE BASED ON THE EXTENSION/ADAPTION OF THE DENEB-DVEGA ARCHITECTURE SO AS TO MIDDLEWARE COULD BE ¿VIRTUALIZED¿, ALLOWING PROCESSING UNITS TO BE ADDED OR REMOVED IN A DYNAMIC WAY, IN ORDER TO ADAPT TO THE CHANGING ENVIRONMENT,IN THE CONTEXT OF SCIENTIFIC WORKFLOWS, LARGE, COMPLEX AND HETEROGENEOUS DATA ARE OPERATED BY LONG RUNNING TASKS THAT MUST BE EXECUTED ON A GREAT NUMBER OF DISTRIBUTED AND HETEROGENEOUS RESOURCES, THEREFORE, WORKFLOW MANAGEMENT TOOLS SHOULD BE ABLE TO PREDICT, AS A PREVIOUS STEP TO ITS DEPLOYMENT, IF A GIVEN WORKFLOW WILL BE CORRECTLY EXECUTED AND IF THE EXPECTED PERFORMANCES WILL BE ACHIEVED, IT WILL BE AN IMPORTANT OBJECTIVE OF THE PROJECT TO PROPOSE (OR TO ADAPT) NEW VERSIONS OF ANALYSIS TECHNIQUES FOR VALIDATING THE CORRECTNESS OF EXISTING SCIENTIFIC WORKFLOWS, THE ANALYSIS TECHNIQUES WILL RELY ON THE UNDERLYING OBJECT NET MODEL, SO THAT MODEL CHECKING TECHNIQUES OR PETRI-NET ANALYSIS TECHNIQUES CAN BE APPLIED, THE PROJECT WILL NOT ONLY CONCENTRATE ON BEHAVIOURAL PROPERTIES OF THE WORKFLOWS, BUT ALSO IN PROVIDING WITH SOME METRICS AND MODEL SIMULATION TECHNIQUES IN ORDER TO PREDICT THE PERFORMANCES OF WORKFLOWS WHEN ARE EXECUTED ON DIFFERENT PARALLEL ENVIRONMENTS AND ASSIST TO USERS TO IMPROVE THE EXECUTION BEHAVIOUR OF WORKFLOWS FROM A PERFORMANCE POINT OF VIEW,