[Paper Review] CBC Approach for Evaluating Potential SaaS on the Cloud

Dhanamma Jagli, Sunita Mahajan+1 more|arXiv (Cornell University)|May 25, 2019

Cloud Computing and Resource Management6 references4 citations

TL;DR

This paper proposes a Constraint-Based Clustering (CBC) model to evaluate potential Software as a Service (SaaS) offerings in cloud environments by integrating both user requirements and quality attributes. The approach enhances traditional SaaS evaluation by clustering services based on functional and non-functional criteria, improving alignment between service capabilities and user needs.

ABSTRACT

The cloud computing is evolving as a key computing platform for sharing resources like infrastructure, platform, software etc. This has proven to be an essential requirement for extending many existing applications. Software as a service (SaaS) is referred as on-demand software supplied by service providers in which software and associated data are hosted on the cloud and it can be accessed by service users using a thin client via a web browser. SaaS is commonly utilized and it provides many benefits to service users. To realize these benefits, it is essential to evaluate potential quality of SaaS, not only to the service users but also to the service providers. They have to evaluate their services against requirements of service users. The existing evaluation models are focusing only on quality attributes of SaaS. In this paper, a new evaluation model is proposed based on the data mining technique of Constraint Based Clustering (CBC). The proposed model gives emphasis on potential requirements of service users along with quality attributes of services.

Motivation & Objective

• To address the gap in existing SaaS evaluation models that focus only on quality attributes, not on user-specific requirements.
• To develop a comprehensive evaluation framework that considers both functional needs and non-functional quality attributes of SaaS.
• To improve service selection accuracy by aligning SaaS offerings with actual user expectations and constraints.
• To enable service providers to benchmark their offerings against user demands using data-driven clustering.
• To enhance decision-making in cloud SaaS adoption through a hybrid evaluation model grounded in data mining techniques.

Proposed method

• The proposed model uses Constraint-Based Clustering (CBC), a data mining technique that groups SaaS services based on predefined constraints.
• Constraints are derived from user requirements (functional) and quality attributes (non-functional), such as performance, security, and availability.
• The clustering process prioritizes services that satisfy both user-specific constraints and quality-of-service (QoS) criteria.
• The model integrates user input and service metadata to form a multi-dimensional evaluation space for clustering.
• The resulting clusters represent groups of SaaS solutions with similar capability profiles, enabling targeted selection.
• The approach supports both service users and providers in identifying optimal SaaS matches through structured, constraint-driven analysis.

Experimental results

Research questions

• RQ1 How can SaaS evaluation models be enhanced to include not only quality attributes but also specific user requirements?
• RQ2 To what extent can Constraint-Based Clustering improve the accuracy of SaaS recommendation and selection?
• RQ3 What are the key constraints that differentiate high-potential SaaS offerings from others in a cloud environment?
• RQ4 How does the integration of functional and non-functional criteria affect the clustering outcome of SaaS services?
• RQ5 Can CBC-based evaluation support both service users and providers in making informed decisions about SaaS adoption and improvement?

Key findings

• The CBC-based model successfully clusters SaaS services by aligning them with both user requirements and quality attributes.
• The approach improves service selection by identifying SaaS solutions that best match user-defined constraints.
• The model enables service providers to benchmark their offerings against user expectations, supporting continuous improvement.
• The integration of functional and non-functional criteria enhances the precision of SaaS evaluation beyond traditional quality-only models.
• The evaluation framework demonstrates practical applicability in real-world SaaS selection scenarios, as validated through the proposed case study.
• The use of data mining via CBC allows for scalable and automated SaaS evaluation in dynamic cloud environments.