A DSL-based Approach for Elasticity Testing of Cloud Systems

Michel Albonico (1, 2) , Amine Benelallam (1, 2) , Jean-Marie Mottu (1, 2) , Gerson Sunyé (1, 2)

1 AtlanModels - Modeling Technologies for Software Production, Operation, and Evolution
2 LINA - Laboratoire d'Informatique de Nantes Atlantique

Elastic cloud systems automatically respond to workload changes by (de-)allocating resources according to a cong-uration specication. Testing such systems requires eort from the tester. In particular, the tester is brought to specify the sequence of resource variations he/she is willing to test the system under, and then, drive the system through this specic sequence of variations while the test is being executed. In this paper, we propose a Domain Specic Language (DSL) aiming at reducing the tester's eort to write and execute elasticity testing. Our DSL abstracts test case specication from the cloud provider's libraries, making our language portable. The DSL compiles into executable code that implicitly orchestrates the execution of elasticity testing. In our preliminary results, we show that our approach reduces the amount of words to specify test cases w.r.t. dedicated libraries. It also shows how much this improvement scales when running a test on multiple cloud providers.

Domaines: Modélisation et simulation; Langage de programmation [cs.PL]

Michel Albonico, Amine Benelallam, Jean-Marie Mottu, Gerson Sunyé. A DSL-based Approach for Elasticity Testing of Cloud Systems. Domain-Specific Model Workshop, Oct 2016, Amsterdam, Netherlands. ⟨10.1145/3023147.3023149⟩. ⟨hal-01437137⟩