TL;DR
This paper introduces a novel 2-pronged framework to compare and analyze existing cloud service brokerage solutions. It addresses the growing complexity of cloud architectures by providing a structured way to understand and evaluate these solutions, highlighting key challenges and research directions for future development.
What Problem Does This Paper Address?
The rapid evolution of cloud computing has led to a proliferation of services and architectures, making it increasingly difficult for organizations to effectively manage, integrate, and leverage these resources. Cloud service brokerage has emerged as a critical concept to address this complexity, acting as an intermediary that mediates between cloud providers and consumers. However, a lack of standardized frameworks for comparing these brokerage solutions hinders research and development in this area. This paper aims to fill this gap by providing a comprehensive comparison framework and a review of existing solutions, identifying architectural concerns and technical requirements.
Methods
The authors propose a two-pronged comparison framework:
- Basic Classification: A categorization schema to broadly classify solutions based on their system type (e.g., IaaS Fabric Controller, API Library, PaaS Solution) and distribution model (open-source vs. proprietary). This is presented in Tables 1 and 2.
- Descriptive Classification: A more detailed, layered taxonomy that examines solutions across three key facets:
- Architecture & Interoperability: Focusing on the underlying architecture and how solutions interact with existing cloud infrastructure and services.
- Languages & Programming: Analyzing the service description languages, programming models, and service engineering aspects.
- Quality: Evaluating aspects like scalability, elasticity, and Quality of Service (QoS)/Service Level Agreements (SLAs).
This descriptive classification is further detailed in Tables 3, 4, and 5. The paper reviews a selection of open-source solutions and publicly funded frameworks to apply this framework.
Datasets
Not discussed in the paper. The paper focuses on reviewing existing open-source solutions and frameworks rather than introducing new datasets for training or evaluation.
Key Insights and Ideas
- Emergence of Brokerage on Top of Management: The research highlights a trend where cloud brokerage solutions are increasingly built upon existing cloud management platforms, indicating a natural evolution towards more sophisticated service orchestration.
- Need for Brokerage and Marketplace Separation: The paper distinguishes between the needs of cloud brokerage (automation, matching requirements to capacity) and cloud marketplaces (bringing providers and customers together, often in an appstore model), suggesting that while related, they address different core functions.
- Commoditization as a Key Enabler: The concept of "commoditization" is presented as crucial for brokerage. This involves making services at all layers (IaaS, PaaS, SaaS) uniformly processable for selection, adaptation, integration, and aggregation, paving the way for more seamless integration and marketplace functionality.
- Interoperability is Paramount: The need for interoperability is consistently emphasized, especially in the context of federated clouds and the desire for a "total commoditization of IaaS" where any compute resource can be plugged into a user's capacity.
- Beyond Performance: Data Integrity and Security: While performance is a common matching criterion, the paper points out that data integrity and security policy enforcement are often overlooked but critical aspects for cloud interoperability that need formalization.
Key Results
The paper provides a detailed comparison of several open-source cloud brokerage solutions (e.g., OpenNebula, OpenStack, Cloudify, Cloud Foundry, OpenShift, CompatibleOne, 4Caast, Optimis) across various dimensions. While specific quantitative results are not the primary focus, the paper's key result is the structured framework itself and the insights derived from applying it:
- Categorization of Solutions: Tables 1 and 2 offer a clear categorization of open-source cloud systems by type and core capabilities.
- Detailed Feature Comparison: Tables 3, 4, and 5 provide a granular comparison of solutions based on their architecture, programming models, and quality attributes, allowing for a nuanced understanding of their strengths and weaknesses.
- Identification of Trends: The review reveals trends such as the move from provider-oriented to developer-oriented and then to end-user-oriented cloud management, and the increasing importance of service description mechanisms for abstraction and composition.
Limitations and Future Work
Limitations:
- The paper primarily focuses on open-source solutions, potentially overlooking the full landscape of commercial brokerage offerings.
- While the framework is comprehensive, the application of the descriptive classification is selective, focusing on the "most advanced ones for each aspect" rather than a complete deep dive into all reviewed products.
Future Work:
- Formalizing Non-Functional Requirements: Developing formal languages to describe common aspects of data integration, security policy enforcement, and SLA management to improve cloud interoperability.
- Cloud Capacity Clearing-Houses: Investigating the potential for cloud capacity clearing-houses that operate like spot markets for buying and selling spare cloud capacity.
- Exploring Alternative Marketplace Models: Moving beyond the appstore model to explore other forms of online marketplaces suitable for cloud apps and their composition.
- Trust Mechanisms: Developing mechanisms for vetting individual providers and enabling trust in layered, federated, and brokered cloud solutions.
- Federated Cloud Architectures: Further research into reference architectures, scope of control, and syndication for distributed cloud architectures.
Summary updated