From KAREN wiki
Contacts
KAREN champion
Robert Gibb
- Geospatial Informatics Research Leader
- Email: GibbR AT LandcareResearch DOT co DOT nz
- Blog: Robert's blog for discussions on SCENZ-Grid & workflows
Landcare Research Project Plan
Additional contacts
Dr Jerry Cooper
- Informatics Science Leader
Greg Jones
- Manager Information Services & Knowledge Management
Number of projects using KAREN
KAREN Projects
SCENZ-Grid
SCENZ-Grid, a collaboration between Landcare Research and GNS Science, is building a GRID geospatial infrastructure on KAREN. SCENZ-Grid is assembling the building blocks for:
- a Science Collaboration Environment and
- a Spatial Computation Engine.
The Science Collaboration Environment will be a portal with access to managed geoprocessing workflows and spatial data, and the Spatial Computation Engine will provide one or more clusters to execute the workflows. Initially the primary cluster will be at Landcare Research in Palmerston North with 104 compute cores, 0.4TB RAM, 1.6TB /tmp and access to 15TB of storage.
SCENZ-Grid is also collaborating with:
- BestGrid at Auckland University
- SEEGRID/Auscope in Perth, Australia
- GEONGRID in USA
We are exploring a range of initial workflows during the 2008/09 research year including:
- Lithology - comparing GNS and Landcare's data about the upper crust between the bottom of the soil profile and top of the rock strata,
- national possum population modelling, and
- on-demand LiDAR & DEM processing
If you have a geospatial problem that might benefit from having access to significantly larger data storage and/or computational capabilities than you normally have access to, then drop an email to Robert Gibb (see above) so that we can explore the possibilities offered by SCENZ-Grid.
Institutional Capability
- Landcare Research's offices in Lincoln, Palmerston North, Hamilton & Tamaki are connected to KAREN for routine Intranet traffic and there is a schedule to connect the Dunedin Office
- In conjunction with Crop & Food, we are holding a series of KAREN workshops to explore ways that we can introduce new KAREN capabilities to a wider science audience
KAREN Workshops
Two joint Landcare and Crop & Food workshops have been held, the first in Lincoln & second in Palmerston North.
Notes from the Lincoln workshop on Wednesday August 27 are at Crop_&_Food_Research#Live_Feedback_from_KAREN_Workshops on Crop & Food's wiki page.
Notes from the joint workshop in Palmerston North about KAREN with representatives from Landcare Research, Crop & Food, and HortResearch. Held October 29, 2008.
Opportunities to Use KAREN
- Enhanced collaboration with remote sites, eg, shared applications, and video/voice collaboration
- Learning new ways of collaborating ... Eg, availability / presence to signal that you're available to meet
- Sharing of large datasets across organisations, with a common point to access and locate ("it's invisible")
- Extend and reach of monitoring technologies
- Ability to understand data once people have left an organisation
- Access to large datasets, for remote analysis (without bringing down to desktop). Brings standardisation issues.
- Needs "new ways of thinking" about sharing
- Needs "new ways of publishing" (about findings AND the raw data)
- Environmental modelling networks ... change from manually getting data from sensors to the use of communications networks
- Greater processing power for uncertainty analysis or sensitivity analysis, eg, Monte Carlo simulations, "run it 10,000 times" to model uncertainty
Organisational Issues
Crop & Food Research
- What do we have? (an audit of what's available)
- Scientists are too busy ... Very hard to get them to come to meetings
- Getting buy-in for people
- They have to see value in it
- Priorities ... Where does this fit in?
- "A big pipe?" Who cares. Need to go beyond this, to applications.
- Unaware that you could be working better
- Lack of access to KAREN at your desktop (but ... It's easy to get it there ... What are you going to do with it once it's there?)
- Time to adapt to a new way of working.
- "How KAREN changed my life" ... Need good stories about how people are doing things
- For the Crop & Food and HortResearch merger ... Ability to bring the people together is critical. Eg, something like presence and availability to signal that you're available
- EVO is a lot better than it used to be, but it's not friendly enough ... (Interface needs to be simpler, Especially for infrequent usage)
- Requires a change in thinking about sharing raw data ... many implicit factors, requires common mental models to enable share; quality control issues
- Higher-level issues:
- Where does this fit with our strategy?
- How do we prove the benefit? It's so new. (Need time to go, before we can measure)
- Concentrate on finding the early adopters (are they aligned with the strategy?)
- "Kick back and let the Gen-Ys take some risks", compared with "asking people who have been doing things the same way for 15 years to change the way they do science."
Landcare Research
- Computationally enabling - Can relax about the speed of transfer, but barrier for use of computational processing, how do you design models to run on super computers. Unsure about how much work would be required and what would we have to do? For a lot of scientists they won’t even be able to conceptualise what the barriers might be. Some of us know what we don’t know, but others don’t.
- If I’m running it in MatLab now, will that work? Do I have to transfer it entirely?
- Demand on IT skills - Much greater demand therefore on the scientists who have IT skills. So need a funding system that accepts that there’s a cost to adding this capability, in particular an enabling short term activity/hump of resources to get things going.
- Intellectual Property - Also, the IP/commercialisation issues. We want to share data across organisations but what are the issues in terms of IP ownership.
- Demands at point of publication about what data is made available to a wider audience, and how that is released
- Doing all of this in a way that’s usable to people other than just the early adopters (and the lawyers)
- Costs - If you want to access a supercomputer, is there cost involved, how do we pay for it? Lack of information/knowledge for this. Communication constraints about letting people know what’s there.
- Specialist tools - Are there the specialist tools already built to do the things I need to do. Are the cloud versions already there, do I have to cloud enable them?
- Green footprint - What is the green footprint, greenness of the high performance computing. What’s the carbon cost, what’s the energy use, can we understand and work this out so we’re not green in every area but not here.
- Financial issues - Financial issue within the organisation in terms of cost/benefit of belonging to KAREN, trading this against future benefit. Tension between operational side and science side.
Next Actions
- We need to build something that's really cool and interesting
- Quickest way ... "allow the Gen-Ys to get going" ... And they'll pull the oldies along
- Force people to use it, eg, by putting meeting notes on Sakai
- Give them a grace period to learn
- Use KAREN logos ("powered by KAREN") T-shirts?
- Build clear deliverables between science and IT
- What are the steps?
- Get science managers to think about the longer term things
- "Design patterns" ... What would research patterns look like in a KAREN environment?
- Would love to have a full day workshop with the modelling team
- Will probably run a workshop in Auckland with Landcare and HortResearch
