TL;DR
This paper proposes a conceptual framework for cloud service brokerage, focusing on a standardized service description template and a reference architecture. The goal is to enable cloud services to be treated as commoditized utilities by providing a structured way to describe and integrate them across different cloud layers (IaaS, PaaS, SaaS).
What Problem Does This Paper Address?
The increasing complexity and diversity of cloud services make it challenging for organizations to effectively manage, integrate, and leverage them as flexible, on-demand utilities. Cloud brokers are emerging as key intermediaries, but they require a robust mechanism to understand, compare, and orchestrate services from various providers. Existing service description approaches, like WSDL for web services, often fall short in addressing the specific needs of cloud environments, particularly in enabling cross-layer integration (from infrastructure to application) and facilitating automated provisioning and management. This paper aims to bridge this gap by defining a comprehensive service description template and a supporting reference architecture.
Methods
The authors propose a two-part conceptual framework:
- Reference Architecture for Cloud Brokers: This architecture extends the NIST cloud architecture specification to explicitly address the challenges of horizontal (across different providers) and vertical (across IaaS, PaaS, SaaS layers) integration. It outlines the roles of a cloud broker in service intermediation, aggregation, and arbitrage.
- Service Description Template: This is the core contribution. It's a structured template designed to describe mediated and integrated cloud services. The template is organized into four main parts:
- Operation: Defines the functional service interface, drawing inspiration from WSDL.
- Quality: Captures non-functional properties like availability and response time.
- Resources: Details deployment requirements, including workload considerations.
- Policies: Encompasses business agreements, security, privacy, compliance, and governance constraints.
To formalize this template, the authors propose using an ontology based on the Web Ontology Language (OWL). This ontology provides a hierarchical structure and allows for reasoning and interoperability. They also suggest a set of sublanguages (definition, configuration, constraints, composition, manipulation) to support different actors and processes within the brokerage framework.
Datasets
Not discussed in the paper.
Key Insights and Ideas
- Bridging the Cloud Layers: A significant challenge in cloud brokerage is achieving seamless integration across the IaaS, PaaS, and SaaS layers. The proposed reference architecture and template explicitly aim to address this vertical interoperability.
- Service as a Commodity: The ultimate goal is to treat cloud services as commoditized utilities, akin to traditional utilities. This requires standardized, comprehensive descriptions that allow for automated discovery, selection, and management.
- Ontology-Driven Formalization: Using OWL for ontology development provides a formal, machine-readable way to define service descriptions, enabling reasoning, validation, and interoperability.
- Multi-faceted Service Description: A truly effective service description needs to go beyond just functional aspects (operations) to include critical non-functional properties (quality), deployment needs (resources), and business/security constraints (policies).
- Layered Language Approach: Recognizing that different stakeholders (developers, operators, users) have different needs, the paper proposes a suite of specialized languages derived from the core ontology to facilitate interaction with the brokerage system.
Key Results
The paper presents a conceptual model and a formalization approach for service description templates. While not reporting on empirical results from a deployed system in detail, it outlines an implementation architecture using OpenStack (IaaS) and CompatibleOne (PaaS broker) as a testbed to validate the proposed framework. The authors argue that their approach, by leveraging ontologies and a structured template, addresses the limitations of existing solutions and provides a foundation for robust cloud brokerage.
Limitations and Future Work
- Implementation and Validation: While an architecture is proposed, the paper acknowledges that the full implementation and validation of the service template model within the testbed architecture are ongoing.
- Operator Calculus Specification: The theoretical work on the operator calculus for architecture manipulation needs to be fully completed.
- Language Support: The paper suggests a range of languages but notes that their full specification and integration require further work.
- Dynamic Aspects: While the template covers static descriptions, the dynamic aspects of service management and adaptation in real-time scenarios could be further explored.
The authors suggest that their proposed OpenStack and CompatibleOne architecture will serve as a basis for developing a solution based on the defined languages for service templates, which remains future work.
Summary updated