1. Do we have our heads in the cloud? THE US NATIONAL VIRTUAL OBSERVATORY
Matthew J. Graham (Caltech, NVO)
IVOA Trieste: Grid & Web Services/AstroRG
20 May 2008

2. IVOA Trieste: Grid & Web Services/AstroRG
The Cloud is…
An elephant in a room of blind men:
SaaS
Utility computing
Web services
Platform as a service
Managed service provider
Service commerce platform
Internet integration
All those things that the Grid promised but turned out not to be:
fast
scalable
resilient
available
A virtualised notion of distributed computation, storage and communication over unlimited resources
IVOA Trieste: Grid & Web Services/AstroRG
20 May 2008

3. IVOA Trieste: Grid & Web Services/AstroRG
Virtualisation
Virtual workspaces:
An abstraction of an execution environment that can be made dynamically available to authorized clients by using well-defined protocols
resource quota (e.g. CPU, memory share)
software configuration (e.g. O/S, provided services)
Implement on Virtual Machines (VMs):
abstraction of a physical host machine
hypervisor intercepts and emulates instructions from VMs and allows management of VMs
VMWare, Xen, etc.
Provide infrastructure API
Plug-ins to hardware/support structures
IVOA Trieste: Grid & Web Services/AstroRG
20 May 2008

4. Different types of cloud
Amazon:
“computing in the cloud”
EC2 (with S3, SQS and SimpleDB)
Use Xen VMs
Images, Instances and Security
Google:
Apps: Python module/API (with BigTable)
Working also with IBM (Blue Cloud)
Microsoft:
Windows Live
Silverlight
Yahoo:
Pipes
Working with Computational Research Laboratories (CRL)
Sun
Project Hydrazine (JavaFX)
Zoho
Adobe (AIR)
Commercial providers:
SalesForce
FlexiScale from XCalibre/Virtual Iron: “cooler than EC2”
Problem becomes availability of the right type of environment (hypervisor)
IVOA Trieste: Grid & Web Services/AstroRG
20 May 2008

5. Globus Workspace Abstracts hyperviser using WSRF
Metadata
instance
Deployment
request
Workspace
Factory Service
Service
Workspace Resource Instance
Deployment RP
Inspect & manage
Notify
Create()
Authorize
& instantiate
Abstracts hyperviser using WSRF
Back-end implementation using Xen VMs on a variety of “science clouds”
Contains GLOBUS
IVOA Trieste: Grid & Web Services/AstroRG
20 May 2008

6. IVOA Trieste: Grid & Web Services/AstroRG
Clouds live in the Sky
A lightweight abstraction layer a la VOSpace
Resource oriented
Data and services both treated as first-class entities
Uses UWS interface for resource management
IVOA Trieste: Grid & Web Services/AstroRG
20 May 2008

7. Identifiers and representations
Use XRIs to identify all resources including computational results:
Different representations (in RDF) returned dependent on resource aspect:
Aspect
Meaning
Representation
object
Data object
VOSpace <node>
image
Service image
VOResource
instance
Service instance
UWS metadata
IVOA Trieste: Grid & Web Services/AstroRG
20 May 2008

8. IVOA Trieste: Grid & Web Services/AstroRG
Use case
I have a data analysis service:
ivo://nvo.caltech/MyService
Upload bytes of a VM image:
Create service instance:
HTTP POST to
Returns
Commit instance to execution:
HTTP POST to
Define computation:
Link to results:
Retrieve bytes:
IVOA Trieste: Grid & Web Services/AstroRG
20 May 2008

9. Is this just blue sky thinking?
“There’s probably a future business in offering an abstraction layer between cloud computing services”
“The missing element is cloud abstraction”
Proof-of-concept system based on VOSpace, UWS and a resource-oriented middleware layer in front of EC2 (and Google Apps)
IVOA Trieste: Grid & Web Services/AstroRG
20 May 2008