TY - CHAP
T1 - Integrated Timed Architectural Modeling/Execution Language
AU - Bacchiani, Lorenzo
AU - Bravetti, Mario
AU - Giallorenzo, Saverio
AU - Mauro, Jacopo
AU - Zavattaro, Gianluigi
PY - 2024
Y1 - 2024
N2 - We discuss an integrated approach for the design, specification, automatic deployment and simulation of microservice-based applications based on the ABS language. In particular, the integration of architectural modeling inspired by TOSCA (component types/port dependencies/architectural invariants) into the ABS language (static and dynamic aspects of ABS, including component properties, e.g., speed, and their use in timed/probabilistic simulations) via dedicated annotations. This is realized by the integration of the ABS toolchain with a dedicated tool, called Timed SmartDepl. Such a tool, at ABS code compile time, solves (starting from the provided architectural specification) the optimal deployment problem and produces ABS deployment orchestrations to be used in the context of timed simulations. Moreover, the potentialities and the expressive power of this approach are confirmed by further integration with external tools, e.g.: the Zephyrus tool, used by Timed SmartDepl to solve the optimal deployment problem via constraint solving, and a machine learning-based predictive module, that generates in advance data to be used in a timed ABS simulation exploiting such predicted data (e.g., simulating the usage, during the day, of predicted data generated during the preceding night).
AB - We discuss an integrated approach for the design, specification, automatic deployment and simulation of microservice-based applications based on the ABS language. In particular, the integration of architectural modeling inspired by TOSCA (component types/port dependencies/architectural invariants) into the ABS language (static and dynamic aspects of ABS, including component properties, e.g., speed, and their use in timed/probabilistic simulations) via dedicated annotations. This is realized by the integration of the ABS toolchain with a dedicated tool, called Timed SmartDepl. Such a tool, at ABS code compile time, solves (starting from the provided architectural specification) the optimal deployment problem and produces ABS deployment orchestrations to be used in the context of timed simulations. Moreover, the potentialities and the expressive power of this approach are confirmed by further integration with external tools, e.g.: the Zephyrus tool, used by Timed SmartDepl to solve the optimal deployment problem via constraint solving, and a machine learning-based predictive module, that generates in advance data to be used in a timed ABS simulation exploiting such predicted data (e.g., simulating the usage, during the day, of predicted data generated during the preceding night).
U2 - 10.1007/978-3-031-51060-1_7
DO - 10.1007/978-3-031-51060-1_7
M3 - Book chapter
AN - SCOPUS:85184284855
SN - 978-3-031-51059-5
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 169
EP - 198
BT - Active Object Languages: Current Research Trends
A2 - de Boer, Frank
A2 - Damiani, Ferruccio
A2 - Hähnle, Reiner
A2 - Johnsen, Einar Broch
A2 - Kamburjan, Eduard
PB - Springer
ER -