MASADA: A Modeling and Simulation Automated Data Analysis Framework for Continuous Data-Intensive Validation of Simulation Models

RODRIGO CASTRO; MATIAS BONAVENTURA; DANIEL FOGUELMAN

Las Palmas

Conferencia; 30th European Simulation and Modelling Conference - ESM'2016, October 26-28, 2016, SIANI, Univ. of Las Palmas; 2016

Complex networked computer systems are subjected toupgrades on a continuous basis. Modeling and simulationof such systems helps with guiding their engineeringprocesses when testing design options on the real systemis not an option. Too often many systems? operationalconditions need to be assumed in order to focus on thequestions at hand, a typical case being the exogenousworkload. Meanwhile, soaring amounts of monitoring informationis logged to analyze the system?s performancein search for improvement opportunities. Concurrently,research questions mutate as operational conditions varythroughout its lifetime. This context poses many challengesto assess the validity of simulation models. Asthe empirical knowledge base of the system grows, thequestion arises whether a simulation model that wasonce deemed valid could be invalidated in the context ofunprecedented operation conditions.In this work we present a conceptual framework and apractical prototype that helps with answering this questionin a systematic, automated way. MASADA parsesrecorded operation intervals and automatically parameterizes,launches, and validates simulation experiments.We tested MASADA in the data acquisition network ofthe ATLAS particle physics experiment at CERN. Theresult is an efficient framework for validating our modelson a continuous basis as new particle collisions imposeunpredictable network workloads.