Modular models for systems based on multi-tenant services: A multi-level petri-net-based approach

Abstract

Recently, the emergence of multi-tenant services-based systems (SBS) in cloud has gained significant attention. These systems, commonly adopted by Cloud software vendors, offer the advantage of building software as a service (SaaS) in a multi-tenant environment with distinct variability plans. However, the task of managing the SBS is challenging due to the interdependence among its diverse components. The difficulty primarily stems from the specific performance requirements of different variability plans and the interdependence between these plans. The main concerns involve effectively representing the components and levels of this system, and establishing synchronization rules to ensure its seamless operation. Refining these rules is crucial for improving coordination among the various components and levels of the SBS. However, it becomes particularly challenging when there are correlation issue among services that fall under different plan characteristics. In this context, it is vital to guarantee performance differentiation for each tenant by adhering to predefined quality of service (QoS) specified in Service Level Agreements (SLAs).To address these challenges, we propose a modular modeling approach based on multi-level Petri-Nets (n-NLS). By adopting these approach, the ambiguity of SBS system can be significantly reduced. An example of a SBS application illustrates our modular modeling approach using n-NLS.