Toward a Unified Ontology of
Cloud Computing
Lamia Youseff Maria Butrico,
University of California, Dilma Da Silva
Santa Barbara IBM T.J. Watson Research Center
1
In the Cloud
• Several Public Cloud Computing Offerings
* Nimbus @ University of Chicago
* Stratus @ University of Florida
2
In the Cloud
Service Oriented Architectures (SOA)
Hardware as a service (HAAS)
Hardware-assisted Virtualization
Paravirtualization Peer-to-Peer Computing
Software as a service (SAAS)
Distributed and Grid Computing
Map-Reduce Data as a service (DAAS)
3
Goal of our study
• Understanding the cloud computing
landscape
– Dissection of cloud computing field
• Five main layers
– Depict inter-relationship between the layers
– Depict inter-dependency on preceding
technologies
• Our Contribution
4
Outline
• Introduction: In the cloud
• Goal of our study
• Motivation
• Classification Methodology
• Cloud Ontology
• Research Opportunities and Discussions
• Questions
5
Classification Methodology
• Principle of Composibility (SOA)
• Stack of layers
– One Cloud layer is higher in stack
• If its services can be composed of the services
of underlying layers.
– Cloud services belong to same layer if they
have equivalent levels of abstraction
• Evident by their targeted users
6
Proposed Cloud Ontology
7
Layer 1: Cloud Application
Software as a Service (SaaS)
• Favorable benefits to its users
• Favorable benefits to its developers
• Construction and Composibility of SaaS.
• Limitations
– Availability of Applications
– Security of Data
– Legacy Application Migration
– Disaster recovery
– Reliable SLA.
8
Cloud S/W
Layer 2:
Environment
• Platform as a service (PaaS)
– Provides APIs and runtime environment
• Favorable benefits to users; i.e developers
• Favorable benefits to the provider
• Example:
open-source map-reduce
Hadoop
9
Layer 3:Cloud S/W
Infrastructure
• Provides fundamental S/W resources
• Composed of three constituents
Computational Resources
Infrastructure as a Service (IaaS)
Enabled by Virtualization
Limitations in Performance Isolation
Data Storage (DaaS)
High Availability, reliability, performance,
replication, data consistency.
Conflicting goals:
– Consistency vs availability
Different Approaches
Communications (CaaS)
Provides network communications with QoS
Research Opportunities:
– Traffic Isolation, Guaranteed message-delay ,
– Dynamic provision of overlays
10
Layer 4: Software Kernel
• Basic software management of Physical servers.
e.g. OS Kernel, hypervisor, VMM, clustering middleware
• Grid and Cluster Computing
e.g. Globus and Condor
• Further integration of Grid Research to Cloud
computing research
– Micro-economics grid models
• For pricing, metering and
supply demand equilibrium.
– Cloud Utility computing
11
Layer 5: Firmware and
Hardware
• Hardware as a Service (HaaS)
– Actual Physical Hardware
• Favorable benefits to users, i.e. Enterprise users
• Favorable benefits to providers
• Challenges :
– Scalable, easy and fast bare-hardware provisioning
– Remote scriptable boot-loaders.
• Example:
IBM Kittyhawk, and
IBM-Morgan Stanley’04 lease
12
Proposed Cloud Ontology
Software as a Service
(SaaS)
Platform as a Service
(PaaS)
Infrastructure
as a Service
(IaaS)
Data Storage Communications
as a Service as a Service
(DaaS) (CaaS)
Hardware
as a Service
(HHaaS)
13
Discussion
• Business incentive for cloud computing
• Pricing models
– Tiered-pricing
– Per-unit pricing
– Subscription based pricing
• Security and Privacy concerns in the cloud
• Monitoring for cloud systems
• Clouds for HPC systems
14
Questions
Lamia Youseff
[email protected]
http://www.cs.ucsb.edu/~lyouseff
15
16
Motivation
• Assist scientific community to expedite
contribution to an emerging field
– Defining Inter-relationship enables enhancing
features.
Example: supporting High Availability and Resilience
Example: Allowing inter-operability between cloud offerings
– Defining Interdependency on preceding
technologies enables identifying limitations and
optimization opportunities
– Simplify the Educational Efforts
17
