NeIC Conference 2013: Report on "Workshop: Infrastructure as a Service for the Life Sciences"

Conclusions from NeIC 2013 workshop IaaS for Life Sciences

There was a generally positive atmosphere to implementing scientific cloud computing along with traditional cluster computing, as this answers many pressing issues for Life Science needs for e-Infrastructure.

Nowadays, a publication of new a scientific life science data analysis method often results the requirement authors to provide the method for others as a tool. If the tool becomes popular, the authors can get in problems as they start to provide "service computing" on their own research infrastructure for other scientists. This is not counter-productive for the goals of life science organisations, as they gain references and visibility this way. e-Infrastructure has an opportunity to increase their impact in life sciences community by increasing support and resources for scientific software tool service activities. Virtual machines hosted on trusted e-Infrastructure providers controlled by life scientist organisation can offer a technical solution to share the workload between life science and e-Infrastructure organisation. This was demonstrated by Nordic ELIXIR node use cases.

Cloud resource taken into use in this way would also benefit from common standards for authentication and authorization services. A lot of discussions revolved on the issue on handling computing over sensitive data. This area contains many unresolved technical and legal issues that involve ICT, for example uniform way to electronically authenticate users, and authorise use of datasets in collaboration with data access committees when access needs to be controlled and audited.

There is a challenge how to get the ICT requirements right. Life science/biomedical community lacks experts who can evaluate if the ICT systems are legally valid for e.g. allowing compute over sensitive data.

Nordic opportunities

• Shared/federated cloud resource between Nordic countries for service computing
• Nordic way of identifying scientific data users (e.g. build on federated identity management and Kalmar Union interfederation)
• Nordic countries aligned in the sense of treating sensitive data

Proposed Nordic Actions

• Promote cloud computing as an alternative for traditonal cluster computing
• Spread information about good scientific cloud iniatives: e.g. extension the (ELIXIR) cloud pilots within Nordic countries. Attract more Life Science use cases
• Certification of cloud service providers on about their service process/security level
• Testing facility for software environments encapsulated in virtual machnine images for compututing consistency
• Organise workshop about cloud service provider security (e.g. intrusion detection)
• Work on legal (and resource allocation) framework for cloud computing across Nordic countries

Group works findings

More than 40 people participated in the showcasing of pilot actions between Finland, Denmark and Norway and about 25 stayed until the very end of the working sessions.

Group 1-2

Pro:

• More efficiently used computational power.
• Freedom -> with responsibility, trust, collaboration
• Scaling
• Elasticity - many service providers to share capacity
• Authentication - Use some common user database CEdugain(?) (joel parse error)
• Cloud hidden behind web application

Cons:

• Cost -> Who is going to pay?
• Possible bureaucracy
• Could you allocate resources fairly?
• Resources Strains(?)(joel parse error) - increase time
• Hard to link a process to a user.
• Sensitive data

Actions:

• Establish cloud as a real alternative
• Share resources between countries

Group 3-4

Strengths:

• Scandinavia aligned legally
• Good collaboration
• High degree of trust
• Strong cloud competence
• Good for resource-poor projects (non-sensitive)
• Potential increase in uptime through redundancy

Weaknesses / Threats:

• Users need to manage images
• Users need help / guidance building images
• Not secure enough for sensitive data

Opportunities:

• Classify users
• Creation / management of images
• Rigid constraints for sensitive data need to be addressed
• Certification for types of data
• Interaction with ethical committee (in REMS context)
• Regression testing / unit testing for images.

Group 5

Strengths:

• Access
• Flexibility / Scalability
• Ease of use
• Standardization
• Resource allocation possibilities

Opportunities:

• VM with bioinfo tool can be moved to where sensitive data and/or large

data reside.

• User-guided development.

Weaknesses:

• Transport of large amounts of data
• Too complicated
• Too slow

Threats:

• Tools that are not used
• Loss of competence
• Legal aspects

Actions:

• Extension of pilots
• NeIC to coordinate and inform about good initiatives

Ulfs additions

"Ws Introduction to IaaS in Life Science in the Nordics"

• The talks focused on showing how great clouds are at reinventing the wheel, while mostly ignoring security problems.
• Elixir have plenty of money but are still in the planning stage.
• Discussion was lame, as most of the audience was dozing off due to the rancid air.

"WS Analysis and Actions"

• Ulf had to leave this, since the rancid air in the auditorium gave me a bad headache.

Session Summary

• Cloud for bioinfo services is nice, but you should not count on the bioinformatics people to be able to keep the resources secure. You need the sysadmin there to do that for you.
• Serving software as VM-images gets the software within the reach of the researchers easily.