부산대학교 도서관

In recent years, the use of mass data processing systems (MDPS) has increased rapidly, both for the processing of research and business data. On the one hand, there is the possibility of buying, pooling, or renting MDPS resources in a scalable and flexible way. On the other hand, there is mass data processing with a certain processing time frame and limited resources to use MDPS resources. As data processing can be mapped to a number of different models of using MDPS resources, it is necessary to be able to assess their adequacy for a given processing, in order to minimize the engagement, i.e. the cost of resources. Such a possibility is provided through processing simulation, which is most often performed using discrete event (DE) simulation. DE simulation of such large systems as MDPS provides challenges in the realization of the simulation itself. Mass data processing of the CERN experiment ALICE Run 3 is, in addition to the large amount of resources required, particularly complex due to the interdependent steps of mass data processing and multi-year duration. Therefore, it is a Long-Term Mass Data Processing (LTMDP). In this paper, an MDPSim simulator for simulating mass data processing will be presented. Finally, a script language for defining a complex workflow in ALICE Run 3 will be described. [ABSTRACT FROM AUTHOR]