Over the past five years, cloud computing has gone from a curiosity to core scientific technology. The cloud's relative simplicity, instant availability, and reasonable cost have made it attractive to scientists, especially in domains relatively new to large scale data analysis. This trend will continue into the foreseeable future, challenging resource providers to adapt their services, to provide easy federation with other providers, and to accommodate many different scientific disciplines. For developers of cloud services, there are also many challenges. Efficient access to, and the curation of large data sets remain largely unsolved problems. Image management also raises new issues, especially if these images are to be shared and trusted. This presentation reviews the current status of cloud computing and presents some ideas on how the upcoming challenges might be met. Presented at CNAF in Bologna, Italy by Charles Loomis in May 2013.

1. Scientific Cloud Computing: Present & Future
Charles (Cal) Loomis (CNRS/LAL & SixSq Sàrl)
INFN CNAF, Bologna, Italy (22 May 2013)

2. 2
Cloud Marketing
“Cloud” is currently very trendy, used everywhere
 Many definitions that are often incompatible
 Used (often) to market pre-existing (non-cloud) software
CommodityComputing(Sun)
UtilityComputing(IBM,HP,…)
AmazonEC2
AmazonEBS
Mature Virtualization
Simple APIs
Excess Capacity

3. 3
In two pages NIST defines:
 Essential characteristics
 Deployment models
 Service models
What is a Cloud?
http://csrc.nist.gov/publications/
nistpubs/800-145/SP800-145.pdf

4. 4
On-demand self-service
 No human intervention
Broad network access
 Fast, reliable remote access
Rapid elasticity
 Scale based on app. needs
Resource pooling
 Multi-tenant sharing
Measured service
 Direct or indirect economic
model with measured use
Essential Characteristics
http://csrc.nist.gov/publications/
nistpubs/800-145/SP800-145.pdf

5. 5
Private
 Single administrative domain,
limited number of users
Community
 Different administrative domains
with common interests & proc.
Public
 People outside of institute’s
administrative domain
Hybrid
 Federation via combination of
other deployment models
Deployment Models
http://csrc.nist.gov/publications/
nistpubs/800-145/SP800-145.pdf

6. 6
Software as a Service (SaaS)
 Direct (scalable) hosting of end
user applications
Platform as a Service (PaaS)
 Framework and infrastructure
for creating web applications
Infrastructure as a Service (IaaS)
 Access to remote virtual
machines with root access
Service Models
http://csrc.nist.gov/publications/
nistpubs/800-145/SP800-145.pdf

7. 7
Advantages
 No software installation
 Universally accessible
Disadvantages
 Questions about data access,
ownership, reliability, etc.
 Integration of services & novel
uses of data are (often) difficult
Trends
 Social scientific computing
 Service APIs to allow
integration  PaaS
Software as a Service (SaaS)

8. 8
Advantages
 Programmers take advantage of
integrated load balancing,
automatic failover, etc.
Disadvantages
 Restricted number of languages
 Applications strongly locked to a
particular provider
Trends
 Dearth of “pure” PaaS offers
 Encroachment from both SaaS
and IaaS sides
Platform as a Service (PaaS)

9. 9
Advantages
 Customized environment with
“root” access
 Easy access to scalable
resources
Disadvantages
 Variety of APIs and interfaces
 VM image creation is difficult
and time-consuming
Trends
 Lots of specialized cloud
providers appearing
 Orchestration pushing into
PaaS space
Infrastructure as a Service (IaaS)

10. 10
Focus: Infrastructure as a Service

11. 11
State of the Art
Commercial Provider: Amazon Web Services (AWS)
 Leading and largest IaaS service provider
 Improving and adding new services at a phenomenal rate
 Almost all IaaS providers use AWS-like service semantics, but
differentiate based on price, SLAs, location, etc.
Commercial Cloud Distribution: VM-ware
 VM-ware: extremely good and complete, but very expensive
 Provide ESXi virtual machine host for free
Open Source Cloud Distributions
 Essentially none in 2004; now easily a dozen different distributions
 StratusLab, WNoDeS, …, OpenStack, OpenNebula, CloudStack
 Very different levels of maturity, stability, scalability, etc.

12. 12
Why are cloud technologies useful?
For (scientific) users
 Custom environment: no rewriting or porting applications to fit into a
resource provider’s environment
 Simple access: most providers use a REST or RPC API allowing
simple access from all programming languages
 Reasonable cost: only pay for what resources are used, especially
attractive for individuals/groups that do not have large, existing
hardware investment

13. 13
Why are cloud technologies useful?
Separation of responsibilities
 Hardware / Services / Platforms / Users
 People at each layer can focus on their responsibilities with minimal
interactions with people in the other layers.
Resource Providers
 Better utilization of shared resource because wider range of
applications (and disciplines) can use the cloud
 With hybrid cloud infrastructures, providers can outsource excess
demand to other providers

14. 14
Trend for Scientific Computing
Will we all just be users of the Amazon cloud?
Pendulum swinging towards large data centers with “fat” machines
 These can offer elastic cloud services at a reasonable price
 With scientific clouds there is low barrier to entry and users can
maintain administrative control of services and data
 Providing shared resource between scientific disciplines much easier
because of virtualization
Migration will be gradual…

15. 15
Overcoming inertia…
Users
 How to use virtual machines to get my work done?
 How to structure, store, access, and protect data?
 Realize shared infrastructures with customized env. are possible
Application Developers
 How to use cloud techniques to improve my applications?
 … and my development workflows?
 Applications can be services (with assoc. pluses and minuses)
Data Centers
 Reuse existing (commodity) hardware investments
 Take advantage of (and train) existing system administrators
 How to manage/use a (private, community, public) cloud?
Significant benefits from cloud even without large scale elasticity!

16. 16
Challenges

17. 17
Elasticity
Can we have infinite elasticity with limited resources?
“Local” solutions
 Economic models to avoid hitting infrastructure limits?
 Spot instances and/or different service classes?
 Aside: IPv6 addressing is necessary for large (scientific) clouds
Federated solutions
 Hybrid (scale-out) infrastructures?
 Higher-level brokers or orchestrators?
Cannot have elasticity without some kind of accounting!

18. 18
Data Management (Legal)
Transfer and treatment of data across borders
 Differing legal protections in different jurisdictions
 Legal constraints for data locality (banking, medical data)
 Unclear responsibilities for data: guardian, custodian, owner, etc.
 Europe working hard to come to a consistent legal framework
Protection of data in the cloud
 Consistent access controls for all data locations
 Guarantees about data protection from cloud provider personnel
 Reliability of the provider’s storage
 Knowledge about provider’s policies for data protection

19. 19
Data Management (Technical)
Efficient exploitation of large datasets
 Need significant computing next to storage, AND/OR
 High bandwidth remote access
Locality matters
 Sometimes it is inconvenient to transfer raw data away from instrument
 Clouds can be used to reduce data locally before transfer
“Open Data” requirements
 Cloud may help meet such requirements
 Does not remove need for well defined dataset metadata and format
 Need long-term funding for the curation of those datasets

20. 20
Security
How to maintain the security of a cloud infrastructure?
Shifting some responsibility to users:
 Users have root access and must secure services within their VMs
 Users have less security experience  need for education & help
 Dynamic network configurations can help improve security
Changing expectations from administrators:
 Leave firewall policies to users running VMs
 Should not expect to run security software inside of VMs
 Need to enhance monitoring to discover abnormal behavior

21. 21
Image Management
Image metadata
 What does an image contain (OS, services, configuration, etc.)?
 What versions of the kernel, software, etc. are included?
 Who is responsible and/or supports a given image?
 How do I identify a given image?
Creating machine images
 How can an image be created for multiple clouds?
 What do I have to do to create a secure machine image?
Sharing images
 How can I make my images available to others?
 Can I parameterize my images to make them useful to more people?
 Can the images be transported and used efficiently?

22. 22
Vendor Lock-in vs. Federation
Common API
 Fully interoperable API avoids duplication in cloud control software
 Has very limited impact on applications and services in the cloud
 Current (quasi-) standards: EC2, OCCI, CIMI, CDMI, …
Common Semantics
 Semantics determine how apps and services operate in the cloud
 For IaaS, cloud providers have a broadly similar semantics, but…
 File-based and block storage is one difference.
Contextualization
 Can a user run the same validated image on all clouds?
 Can a user share the same parameterized image with others?
 Neglected issue in standardization; CloudInit becoming de facto std.

23. 23
Solutions

24. 24
StratusLab History
Informal collaboration to investigate
running grid services on Amazon
EC2 (2007)
StratusLab Project (6/2010 to
5/2012) co-funded by EC with
6 partners from 5 countries
Open collaboration
to continue the
development and support of
the StratusLab software
Website: http://stratuslab.eu
Twitter: @StratusLab
Support: support@stratuslab.eu
Source: http://github.com/StratusLab
Identified need for open
source cloud distribution.
Production dist. with academic
& commercial deployments.

25. 25
StratusLab
Complete Infrastructure as a Service cloud distribution
 Developed within EU project, software maintained by partners
 Focus: Simple to install and simple to use
Services
 Compute: Virtual machine management (currently uses OpenNebula)
 Storage: Volume-based storage service
 Network: Simple configuration for public, local, and private VM access
 Image mgt.: Complete system for trusted sharing of VM images
 Tools (python CLI) and APIs (Libcloud) to facilitate use of cloud
 Tools to facilitate installation of services

26. 26
SlipStream
Cloud orchestrator and deployment engine
 Facilitates testing, deployment, and maintenance of complex systems
 Transparent access
to multiple cloud
infrastructures
 Allows automated
deployment of systems
in one or more clouds

27. 27
Image Management

28. 28
Image Management
Image metadata
 What does an image contain (OS, services, configuration, etc.)?
 What versions of the kernel, software, etc. are included?
 Who is responsible and/or supports a given image?
 How do I identify a given image?
Creating machine images
 How can an image be created for multiple clouds?
 What do I have to do to create a secure machine image?
Sharing images
 How can I make my images available to others?
 Can I parameterize my images to make them useful to more people?
 Can the images be transported and used efficiently?

29. 29
Image Management Actors

30. 30
Marketplace
Priorities
 Mechanism for sharing and trusting images
 Possible to distribute fixed, read-only data sets as well
 Split the storage of image metadata and image contents
 Define roles for creator, user, administrator, and validator
Implementation
 Marketplace API: Proprietary REST API for create, read, search
 Marketplace acts as image registry and handles only metadata
 Image contents can be located on any public (web) server
 ‘Private’ images can also be held in cloud storage

31. 31
Marketplace: Key service in larger ecosystem
Trust
 Marketplace metadata plays key role in providing information about the
image contents and provenance
Factories
 stratus-create-image facilitates customization of images
 VirtualBox and other local virtual tools
 Bitnami and many similar services for build of standard OS services
 Quattor and other fabric management can be used
Transport (& Storage)
 StratusLab uses simple HTTP(S) for image transport
 Could imagine using ftp, gridftp, bittorrent, etc.
 vm-caster/catcher in EGI federated cloud task force

32. 32
Image Handling Workflow

33. 33
Cloud Federation

34. 34
StratusLab Federated Cloud Infrastructure
Features
 Two sites operating (LAL and GRNET) for ~3 years
 Common user authentication
 Ability to use the same images across resources
 StratusLab client allows easy switching between sites
 StratusLab Libcloud binding allows common view of both sites
Need to go further…
 To sites running different cloud software
 Helix Nebula and EGI Fed. Cloud Task Force active in this area

35. 35
Federation Models
Transparent Federation
 Site operators “outsource” to other providers
 Completely transparent to end users
 Difficult to achieve in practice because of data protection concerns and
network access/performance
Brokered Federation
 Variety of different cloud infrastructures are visible to users
 Users choose to place virtual machines in particular locations
 Simple clients can handle federation if differences are small
 Orchestrators are needed for larger differences between clouds
Both Helix Nebula and EGI take the brokered approach

36. 36
Vendor Lock-in vs. Federation
Common API — minor issue
 Fully interoperable API avoids duplication in cloud control software
 Has very limited impact on applications and services in the cloud
 Current (quasi-) standards: EC2, OCCI, CIMI, CDMI, …
Common Semantics — important issue
 Semantics determine how apps and services operate in the cloud
 For IaaS, cloud providers have a broadly similar semantics, but…
 File-based and block storage is one difference.
Contextualization — critical issue
 Can a user run the same validated image on all clouds?
 Can a user share the same parameterized image with others?
 Neglected in standardization but CloudInit becoming de facto std.

37. 37
“Our” Contributions
StratusLab
 Adopt CIMI as the standard interface to services
 Provide Libcloud (python) driver for StratusLab
 Provide EC2, OCCI, etc. as adaptors to CIMI interface
 Move to CloudInit for contextualization framework
 Flexible authentication system to support different methods
SlipStream
 Chosen as main interface to supported clouds in Helix Nebula
 Already supports multi-cloud deployments
 Support for all clouds within Helix Nebula

38. 38
Running Clouds in Production

39. 39
Cloud Experience at LAL
Private cloud for laboratory services
 Works well, plan to migrate all services including grid worker nodes
and experiment-specific servers
 Services switched to VMs without users being aware of change
 Very different way of working, need to change administrator habits
 Have seen some stability issues related to SL6 kernel/virtualization
Public cloud open to university
 Very positive reaction to cloud; LAL resources nearly 100% used
 Variety of disciplines: biology, software eng., statistics, astrophysics,
bioinformatics, …
 After initial introduction, users require only low level of support
 Other labs offering StratusLab training without our direct involvement

40. 40
Summary

41. 41
Summary
Cloud for scientific computing
 Already a common tool for many scientific disciplines
 Cloud technologies will become more pervasive with time
 Associated swing back to larger, centralized data centers
Many challenges for cloud
 Elasticity with limited resources
 Data management (legal & technical)
 Security
 Image management — unique holistic approach from StratusLab
 Federation — brokered federation with StratusLab & SlipStream
LAL cloud experience
 Very positive feedback from both administrators and users

42. 42
Questions and Discussion
Website: http://stratuslab.eu/
Twitter: @StratusLab
Support: support@stratuslab.eu
Source: http://github.com/StratusLab
SixSq: http://sixsq.com
SlipStream: https://github.com/organizations/slipstream

43. http://stratuslab.eu/
Copyright © 2013, Members of the StratusLab collaboration.
This work is licensed under the Creative Commons Attribution 3.0
Unported License (http://creativecommons.org/licenses/by/3.0/).