Source-linked AI summary
Market-Oriented Cloud Computing: Vision, Hype, and Reality for Delivering IT Services as Computing Utilities
Rajkumar Buyya, Chee Shin Yeo, Srikumar Venugopal
TL;DR
The paper addresses how computing can become an on-demand utility and how Clouds can allocate resources through market mechanisms. It defines Cloud computing, proposes SLA-oriented market architecture and resource-management strategies, surveys platforms, and outlines interconnecting Clouds into a global exchange. It concludes that current technologies need negotiation, SLA-aware VM allocation, risk management, and provider interoperability to realize this vision.
Problem
Existing Cloud technologies provide limited support for negotiated QoS, automatic SLA-oriented resource allocation, SLA-violation risk management, and interoperability among providers.
Method
The paper defines Clouds, proposes a market-oriented architecture combining customer-driven service management and computational risk management, surveys platforms, and outlines Cloud markets.
Results
The paper presents an architecture for market-oriented resource allocation within Clouds and a vision for a global Cloud exchange for trading services.
Takeaways & Limitations
Realizing computing as a utility requires extending Cloud systems with SLA negotiation, VM resource allocation, SLA-risk management, interoperability, and supporting market infrastructure.
Takeaways & Limitations
Open Cloud markets face enterprise conservatism, regulatory restrictions on data location, and unresolved legal frameworks for restitution after SLA violations.
Abstract
from arXiv · showhide
This keynote paper: presents a 21st century vision of computing; identifies various computing paradigms promising to deliver the vision of computing utilities; defines Cloud computing and provides the architecture for creating market-oriented Clouds by leveraging technologies such as VMs; provides thoughts on market-based resource management strategies that encompass both customer-driven service management and computational risk management to sustain SLA-oriented resource allocation; presents some representative Cloud platforms especially those developed in industries along with our current work towards realising market-oriented resource allocation of Clouds by leveraging the 3rd generation Aneka enterprise Grid technology; reveals our early thoughts on interconnecting Clouds for dynamically creating an atmospheric computing environment along with pointers to future community research; and concludes with the need for convergence of competing IT paradigms for delivering our 21st century vision.
1. Introduction
The paper frames computing as an on-demand utility and surveys paradigms—including Cloud computing—developed toward that vision. It defines Clouds as virtualized, dynamically provisioned resources delivered under negotiated agreements and notes changing paradigm popularity.
- Computing utilities would provide services on demand, charge users according to access, and reduce the need to build and maintain complex IT infrastructure.
- Cluster, Grid, peer-to-peer, service, market-oriented, and Cloud computing are presented as paradigms pursuing service-based utility computing.
- 1.1 Definition and Trends: A Cloud is a parallel and distributed system of interconnected virtualized computers dynamically provisioned as unified resources under negotiated service-level agreements.
- 1.1 Definition and Trends: Clouds use VM-based virtualization and on-demand provisioning to create personalized resource collections accessible as composable services.
- 1.2 Web Search Trends: Google search trends indicate cluster computing was prominent in the 1990s, Grid computing from the early 2000s, and Cloud computing more recently.
2. Market-Oriented Cloud Architecture
The proposed market-oriented architecture replaces equal-priority, system-centric allocation with SLA-aware resource management. It combines customer service requirements, computational risk, pricing, monitoring, admission control, and VM-based allocation.
- Cloud providers must manage customer-specific QoS requirements negotiated in SLAs rather than treat all service requests as equally important.
- The architecture connects users or brokers with a Cloud through an SLA Resource Allocator that coordinates request examination, admission control, monitoring, pricing, and accounting.
- The Service Request Examiner interprets QoS requirements, consults current resource and workload status, accepts or rejects requests, and assigns accepted requests to VMs.
- VMs can be started and stopped dynamically on one physical machine, enabling flexible resource partitions and isolated operating-system environments.
- Market-oriented resource management combines customer-driven service management and computational risk management to sustain SLA-oriented resource allocation.
- Commercial market-oriented Clouds should incorporate autonomic management and dynamically assign VM resource shares as service requirements change.
3. Emerging Cloud Platforms
The paper surveys representative Cloud platforms spanning infrastructure, application, storage, device-access, and enterprise-Grid offerings. Aneka is highlighted as a configurable service-oriented platform with SLA and negotiation support.
- Industry Cloud offerings include Amazon EC2 and S3, Google App Engine, Microsoft Live Mesh, Sun Grid, and GRIDS Lab Aneka.
- Amazon EC2 provides virtual computing environments for Linux applications, while S3 charges for data transfer and EC2 charges for instance lifetime.
- Google App Engine supports Python Web applications and APIs for datastores, accounts, URL fetching, image manipulation, and email.
- GRIDS Lab Aneka is a .NET-based service-oriented platform whose configurable containers support enterprise Grids, multiple application models, persistence, security, and communication protocols.
- Aneka supports SLA requirements such as deadlines and budgets, with Gridbus enabling remote access and negotiation with providers.
4. Global Cloud Exchange and Markets
The paper proposes a market infrastructure in which standardized Cloud services can be discovered, brokered, priced, and traded under SLAs. It also identifies regulatory, interoperability, and restitution challenges that constrain deployment.
- Market infrastructure: Standard service interfaces would commoditise Cloud services and enable markets for trading computing capacity.A market directory, auctioneers, and banking system support discovery, bid clearing, and financial settlement.
- Market participants: Brokers mediate between consumers and providers by leasing capacity, reselling it, and binding transactions through SLAs.SLAs specify service metrics and penalties for meeting or violating expectations.
- Market participants: Providers can use market-based pricing and capacity planning, while consumers select providers or brokers according to requirements, budgets, and utility functions.Prices may reflect market conditions, demand, and resource utilisation; consumers may trade through advance SLAs or spot purchases.
- Prior work: Existing research systems have already developed market structures and brokers for trading VM-based resource allocations.The paper cites SHARP, Tycoon, Bellagio, Shirako, and Gridbus as examples of enabling technology.
- Challenges: Universal Cloud markets remain constrained by conservative enterprise adoption, proof-of-concept systems, regulatory data-location requirements, and unresolved SLA restitution.The paper identifies legal frameworks for agreements as outside its scope.
5. Summary and Conclusion
The paper presents Cloud computing as a utility paradigm and proposes market-oriented resource allocation and a global Cloud exchange. It concludes that current platforms require extensions for SLA management, risk handling, interoperability, and regulatory support.
- Summary: The paper proposes an architecture for market-oriented Cloud resource allocation and a vision for a global exchange trading services.It also surveys representative Cloud platforms and positions Cloud computing as delivery of IT services as computing utilities.
- Summary: Current Cloud technologies have limited support for QoS negotiation, SLA-based VM allocation, SLA-violation risk management, and provider interoperability.These capabilities are identified as required extensions to state-of-the-art platforms.
- Future directions: A global Cloud exchange would require internetworking platforms with market makers, registries, clearing houses, brokers, payment systems, and accounting infrastructure.The paper also identifies regulatory and legal issues beyond the technical challenges.