Authors:
Khadija Aouzal
1
;
Hatim Hafiddi
2
and
Mohamed Dahchour
1
Affiliations:
1
INPT, Morocco
;
2
INPT and ENSIAS, Morocco
Keyword(s):
SaaS, Non-Functional Variability, SPLE, MDE, QoS Characteristics, SLA.
Related
Ontology
Subjects/Areas/Topics:
Agents
;
Artificial Intelligence
;
Cloud Computing
;
Cross-Feeding between Data and Software Engineering
;
Data Engineering
;
Databases and Data Security
;
Model-Driven Engineering
;
Software Engineering
;
Software Engineering Methods and Techniques
Abstract:
In a multi-tenant architecture of a Software as a Service (SaaS) application, one single instance is shared among
different tenants. However, this architectural style supports only the commonalities among tenants and does
not cope with the variations and the specific context of each tenant. These variations concern either functional
or non-functional properties. In this paper, we deal with non-functional variability in SaaS services in order
to support the different quality levels that a service may have. For that purpose, we propose an approach
that considers Service Level Agreements (SLAs) as Families in terms of Software Product Line Engineering.
We define two metamodels: NFVariability metamodel and VariableSLA metamodel. The first one models
and captures variability in quality attributes of services. The second one models a dynamic and variable
SLA. Model-to-model transformations are performed to transform Feature Model (NFVariability metamodel
instance) to Generic SLA (
VariableSLA instance) in order to dynamically deal with the tenant-specific nonfunctional
requirements.
(More)