Need to know
- System architecture
- projects on system
- dealing with difficult users
- new technologies
- black book of other hpc admin contacts
- personal cell number of vendor
How many FTEs
- Size and Complexity
- N of users'
- support level
- individual responsibilities - admin vs TOEs
- does chief admin require a vacation
- new or esoteric makes you the beta tester
- review needs...only buy what you can use - don't build a system looking for users.
- power and cooling are critical
- sweet spot in pricing - buy more nodes just under top of the line
- commercial software: need to keep track of licensing, maintenance fees
- will software run under your scheduler?
- does your system need an upgrade
- will likely break IB and parallel fs drivers
- will existing research code base need upgrade?
- 2 days to 2 weeks - impact on researchers
Comment: I put a good amount of time into upgrading my cluster. There have been a lot of security breeches.
Response: Need to be careful...there are not a lot of people hacking these clusters. Some patches are not necessary, while others are.
Job scheduling policies
- Keep queue full at all times with a small backlog
- try and schedule whole nodes per job
- long running jobs will KILL throughput
- segregation of nodes with queues for long stuff
- checkpointing can help, but has its own issues
- unique skill set doubles team capapbilities
- what tasks you dislike-hire someone who enjoys those
- complimentary personality and work habits
- good team communication is a must
Conferences: Importance of Social Networking
- Single greatest source of info on what's up
- ?eliminates duplication of effort
- where do you draw the line
- defined in part by n of FTEs
- biggest bang for the buck: time vs productivity
- workshop vs one on one
- code base: replace with something appropriate
keeping the boss happy
- users happy = happy boss
- periodic report on system and user base
- report things that look good on proposal
- filter useless info from good
- + encounters are good
- need institutional support
- awards from NSF, etc.
- condo model to cover costs of hardware
- collaborations with reputable vendors
- high quality research and publications = awards
GPN covers 1.6 the area of Texas with a population that is much less.
GPN's original focus was on networking, but the emphasis now is broader--expanding to CI.
- how to achieve coordination ofefforts
- how to provide services that are valuable
given the demographics mentioned earlier.
How do we leverage
- existing organizational structures?
- Expertise in the gPN member community to meet CI needs behind networking
Started with a CI Advisory Committee leading to two surveys and then a charter for the GPN Program Committee.
CI Program Working Groups
- Education, Outreach and Training
- Divide cost equally
- Pay for use
- Subscription fees
Organizational styles and issues
- Top Down Hierarchical
- Grass Roots up
- Concensus driven, shared objectives
- driven by member needs
- change from inside out
- diverse funding models
Communication and coordination mechanisms are more challenging in grass-roots efforts.
What style is appropriate for GPN CI Program?
Continuing the discussion...Workshop on approaches to regional CI support.
Dana started at OSU as a lecturer in math. She became the Director of the HPC Center. She relied on Henry Neeman and Jeff Pummill to get started.
- Cluster use doubles every year for three years
- Pistol Pete wait times getting longer
- Staff 1.5 FTE over-subscribed
Solution: Write NSF MRI proposal.
- Typically due in late January
- Acquisition proposals are limited to 2 per institution
- must win internal competition, first.
- Read everything on the NSF MRI page - including previous awards
- Study the solicitation
This is an infrastructure proposal which is significantly different from a research proposal.
- Commit to writing. get a few helpful folks to encourage and help
- win pre-proposal for institution MRI slot
- work on full proposal as early as possible
- read solicitation carefully
- nsf program managers want to help
Writing the proposal is
- time consuming
- probability of success is low
- But, it's worth it! If you don't succeed at first, you will get valuable help from reviews.
Big Picture II
- You have to have a need for a supercomputer
- You need the expertise to acquire, deploy and maintain it.
- You need adequate space, power and cooling.
- two page white paper
- required evidence of contact with MRI program manager
- during November info was gathered from potential users
Polling potential users: emailed all current HPC users asking for
- one sentence description of research that will use cluster
- n of users, students, undergrads, women, minorities
- one sentence for broaderimpacts
- I got 25 responses from 20 different academic departments
- Get several quotes
- FY2011 required cost share
- Key Personnel - up to 4 co-PIs, unlimited Senior PErsoinnel. Pick biggest users as co PIs
- Project Description - Make the case to your internal reviewers that your proposal is what the NSF MRI program wants to fund.
The solicitation describes the needed sections
Research Activities Enabled:
Give each research activity/researcher a section. All of this needs to fit into the first 9 pages.
The slides will all be on the web with Dana's comments, as well.
Q: I heard it's easier to get a second award once you receive one.
A: You can say in subsequent proposals that this new thing will improve what you've already funded in this way.
By 2012 80% of all storage capacity sold will be for file-based data.
Traditional Storage Limitations
- many file systems/volumens (islands)
- utilization rates vary across islands
- it takes 640 days to move 1PB across and OC3
Stephen R. Wheat, Sr. Director, HPC Worldwide Business Operations, Data Center Systems Group (DCSG), Intel
Three talks in one:
- message from the sponsor
- who are the missing middle
- making the missing middle not missing
Legal Disclaimer- What I say may or may not lead you to certain actions. We recommend that you use more than this talk to make your decisions.
- The transistor chip is the foundation for all computing.
- Intel has gone from 2% to 80% of the HPC market since 2002.
- In 2003 we predicted the 22nm speed in 2011 and we've hit it. We also predicted 8nm speed by 2017. We are still on track for that.
- People are deploying more and more processors--the systems are becoming 40% bigger. We've seen at least two SCs drop off the list from the first time they were on to the next publication, indicating the rapid change in technology.
- HPC is now about 25% of sales.
- Gone from about $55K per Gflop to <$100 per Gflop while improving performance.
- Leading performance per watt for serial and highly parallel workloads
- Optimized efficiency for a heterogeneous solution in combo with Intel Xeon processors
- Complete set of software tools for scalable solutions
Scaling Programmability: multicore, manycore, cluster (both multi- and many- cores). Not sure whether the cluster setup is in one computer or whether he's talking about a many core cluster within one machine? Sounds like it's in one machine.
- traditional computer users - simple tasks
- Missing Middle - moderately complex tasks
- high-end HPC users - highly complex tasks
The Missing Middle seems to refer to a group of users with moderately complex tasks. They could use HPC, but they do not.
- The lack of application software
- Lack of sufficient talent
- cost constraints
I think the message is that we should be finding those with problems of moderate complexity, get the moderately priced cluster that will fill there need, and help them to use the resources.
These middle users could be small and medium companies in the community that do manufacturing or other types of business that are moderately complex and would benefit from using a few processors.
iRODS - Integrated Rule Oriented Data System, picked up where SRB left off. iRODS has the concept of a unified virtual data collection. The user sees a single collection but the data may be distributed across several locations.
There are global name spaces and a metadata catalog managing mappings between a collection and its data.
It's also to federate among data grids -- a distributed collection of distributed collections.
Highly intensive data sites:
- Genome centers (petabytes)
- NOAA's National Climate Data Center (repository management, publishing public data, support for climate modeling)
- Streamline/Automate data movement
- National/Agency repositories
- Implement management policies - each community defines their own policies
- Automate admin taks
- Validate assessment criteria - verify policy compliance
Additional Design Goals:
- virutalize distributed collections
- abstract out the data management
Policy Implementation - built in rules and triggers
iRODS can be seen as an access interface.
iRODS extensible core infrastructure including
- network transport
- distributed access management
Privacy is a big issue anytime that data hosting and data sharing come into play.
There is health data that is being stored via iROD.
Doug showed a short video of a plane backing up in the air and landing on the deck (successfully) of an aircraft carrier at see.
In 1982 there were 400 or so units sold in 1982 of the Cray XMP/4 ($24M each). The old definition is that a supercomputer was any computer costing over a million dollars.
"For every vision there is an equal and opposite revision."
Parallel computing according to Amdahl's law, 1967, "If I have a series of processors and I parallelize it, how much speedup will I get." Answer was that there's a limit as the n of processors increases. Problem size is generally NOT the fixed parameter but rather the time one is willing to wait for a job to complete. John Gustafson revised the law in 1988: "Gustafson's Law contradicts Amdahl's law, which describes a limit on the speed-up that parallelization can provide." (http://en.wikipedia.org/wiki/Gustafson's_law)
- considered to be the most significant contribution to computational science
- demonstrated that high performance could be achieved through hardware and software scalability
- Question I hear: Why can't I network my desktop PC's to build a supercomputer?
- Well, why can't we give all the undergraduates a pair of binoculars and send them out to the wilderness to study the universe?
How much connectivity is enough, e.g., between nodes in a cluster? 8 connectors, need at least four connections. How do you determine if there is appropriate design for a cluster? Need to consider number of compute nodes and number of connections. This is related to Gustafon's revision of Amdahl's Law.
Able to model the landing of the new plane on the USS Wasp a year before it was physically possible (see opening comment). This modeling was compute intensive--it took into account movement of the ship in addition to movement of the plane.
Scientists are interested in quantification of margins and uncertainties (QMU). Computational Confidence is a concept coming from the weapons side.
Computational Confidence = M/U. (What is M and what is U?) M= Margins and U = Uncertainties.
Drop a bomb off a plane you have both margins and uncertainties.
The Teragrid was run by 11 providers and coordination was difficulty.
The XSEDE solicitation was released in 2008. It was an extremely complicated solicitation. There were two awards for 3 to five years. Also, the former 11 providers become service providers under the new program.
Any user can now get auditing information on N of users, hours, etc.
The largest part of this was to actually integrate services. This was "the big enchilada" or $121M over 5 years. This required National Science Board approval that came in July, 2011.
XSEDE Components: Coordination and Management Service, Advanced User Support Service, Technical Services.
XSEDE is not just about big iron, it's about increased user productivity. It's a distributed facility. There's one award and one PI, John Towns. There's the campus champion program, a portal for single sign on, a help desk, and morel. TACC will reply a new machine in January with 2 PF conventional CPU, 10 PF MiC from Intel.
XSEDE targets US based, open science research.
- Allocate computational and storage resources
- User support
- Support scientific gateways
- Tools to integrate campus and national facilities as well as to enhance inter and intra-campus digital collaborations.
Apparently, one can take the XSEDE tools and use them on campus.
OSG will become a service provider for XSEDE. Globus and Globus Online are elements of XSEDE.
- Collection of stakeholder needs
- assess plans and deployments
- representation in the management of XSEDE
- User advisory committee
- Users on XSEDE board
Q: Is Xsede going to address the issue of heterogeneous systems?
Q: Are there any early success stories?
A: I don't think it's fair to ask since the project is only 3 months old.
Q: What is the plan for data sharing? Is that a tier 2 or 3 type resource? What's the mechanism for moving data to the national infrastructure? Is it a closed or open system?
A: I think that's what we're trying to do.
Discussion of how the EPSCoR office and Henry Neeman were able to coordinate on the RII Track-1, RII Track-2, and RII Cyber Connectivity (C2) proposals that brought several million dollars to Oklahoma. Stressed the importance of organizations like OSCAR and knowledgeable folks like Henry to be able to pull these things off.
OSCER stresses education, first. We are able to cite in our proposals that we have access to a state-of-the-art facility. A facility like OSCER helps with attracting quality faculty, providing an excellent service to us and educating us on the basics of supercomputing and the latest developments.
Henry and I collaborate on the cybercommons project. Henry has helped with our successful MRI proposal. There are other aspects of the OSCER story, including ease of use, that we hope to be able to model.
We are not a research organization. We work on aircraft and maintain the software, other weapons systems and flight simulators. It takes HPC skills to take care of these things.
We need to prepare our workforce, working with GPGPU and parallel processing technologies. We acquire a fair amount of hardware from government re-cycling which is a potential for training and educating our staff.
DoD has huge resources. Not all is tailored to what we need at Tinker so we look to regional and campus HPC centers to help us out.
We can do testing but really can't do production things because it's classified. So OSCER is a useful tool for education and training and assistance in building a cluster. This is a local resource that we can depend on for tours, training and more.
I would give them an A-. The key is service, getting jobs run on the machine, rounds and help sessions are great. The reason for the "-" is the lack of expertise and support for GPGPU.
This is a very interesting presentation on moving large-scale data. I spoke with Ian prior to the presentation and he explained that we would also be hearing from Steve Tuecke about this on Friday (Professional Development seminar). I won't give it away!
Innovation is something that we are all used to. We have a distinguished panel.
RTP (Research Triangle Park) is competing for #3 in the region for innovation dollars. The three reasons we haven't quite achieved everything we can: 1) do we have an adequate ecosystem for producing companies? Not a lot of folks nearby who have created a company. 2) If faculty members are successful at getting research dollars then they are less likely to want to start a company. Our success in that area holds back the development of new startups. 3) There is a shortage of people that you can talk to, informally. Seminars don't work.
At Chapel Hill we are 1) Teaching entrepreneurship to our students. This can be partnered with other degree programs, it seems. 2) Streamlining the bureaucratic hurdle--bureaucrats are good at initiating a rule to prevent something bad from happening. We have an express license with the university that can get done in two weeks. 3) We try to improve the atmosphere where faculty can meet folks who have started a company. We bring in folks to teach a class or two and faculty can meet them, informally.
Discussion of innovation at the state level. People are talking about American innovation sinking. This discussion of innovation is also about the future of our children, the future of our country, aspects of employment and unemployment.
Can we educate our work force? Do we need more college degrees or is it the kind of education that folks receive. Looking at this among states in terms of their abilities to be competitive I have found some indicators that are consistent in explaining business innovation and education innovation.
Business Innovation (most to least predictive):
- % of employment in high tech firms
- % of computer specialists among workforce
- Business R&D among industry
- % of engineers in workforce
- SBIRs as percent of state GDP--
- Venture capital oer million
- s&e grad students per thousand people aged 25 to 34.
Education Innovation (most to least predictive):
- Academic R&D as percent of state GDP
- S&E bachelor's degrees awarded
- Percent of 25 to 44 year olds who are high school grads
- SBIRs per million dollars of state GDP
States who have more students who qualify for LOANS are more likely to be more competitive. States with more students who qualify for PELL GRANTS are less competitive. LOANS are the best single predictor of a state's competitiveness (negative correlation).
We have to help women and minorities understand the roots to success.
Some Controversy: Most of the most successful entrepreneurs DON'T have college degrees.
How did Redhat become a leader: When I joined RedHat in 1996, they were a magazine company placing disks in the back of magazines. They continued to reach out to kids in 1997 & 1998 who understood the Internet but did not have college degrees. These were unbelievably young people who were incredibly constrained by the college setting and that caused them to drop out.
We're here because of Open Source computing -- the Internet is the most visible open source project. Open Source is here to stay!
From 1998 to 2001 our board members tried to push me to move to California from North Carolina. Fortunately, we did not.
The roots of open source are in the academic world. Open source is the content that can fuel education and training and innovation. A person can learn via the MIT coursework that is now online and delve into the roots of open source code.
CIOs of institutions at OMNIPOP asked engineers, "What would we have to do or pay in order to replace the service?"
If the cost is higher than working with OMNIPOP then we are doing the right thing. If not, then we are doing the wrong things.
Just looking at I2, some use OMNIPOP as a primary connection and others as a secondary.
Survey was developed to see which were being used by members. Essential services, value-added services and some that were not being used.
Then, assigned a dollar value to each service.
Handout has average savings per school.
For FY11, they multiplied the average rating for the service by the number of respondents who ranked that item to come up with a value ranking.
The cost of the collaboration was $825K. The value of the return on investment was $3.26 in value for each dollar invested by a school.
The metrics give CIOs confidence that they are on the right path.
For the engineers the answer to the questions, "What if the OMNIPOP went away?" was "We would find/start another collaboration.
Another benefit is that the engineers have evidence that the services they are providing via OMNIPOP are valuable and a good use of time.
This also helps others in CIC to communicate the value this particular collaboration.
Count number of proposals from institutions that indicate they are using the network.
The Q & A were very interesting...
Jerry Gorchow, speaker, I believe...
- Coordinate existing I2 services with new services
- Use external vendors
- Be responsive to member demands
- InCommon is an important piece of this
- Reduced admin process for negotiating prices, contract terms, procurement
This will probably create a lot of buzz in the business world.
Q: Box agreement. Pricing?
A: Tiered based on size of institution. 6 tiers. Tier 1 = 10K accounts, 100T of storage, $27K/Year. (I think that this is the smallest amount).
Q: Will we be able to execute a CAA to be HIPPA compliant?
A: They do not have a BAA(?) so this will have to come later. We will have agreements to assure HIPAA compliance.
Q: What about CMS?
A: Box is more about file share than storage. A 10T dataset from CMS would not be appropriate on Box. The two services are very complimentary (storage and cloud computing) and will work from design of project to collecting data to writing publications.
Q: Well, that raises a follow-up question, how will these be integrated?
A: We would like to brand this as Internet2 and make it possible for current and other vendors to come in behind those services. The key is that we are evolving and others need to become engaged. We need use cases to assist in designing these services.
Q: At Cornell we've been an early adopter of the Box initiative and what we've learned is how rapidly this is being deployed and it represents real progress. These are solutions, now. It allows me to spend less time on this sort of thing and more time on the things that are unique and critical for our campus.
A: Thanks for that.
Q: As a RON I'd like to know how we can help and participate.
A: It makes sense to have some services at a national level and others at the regional level. We haven't teased that apart, yet.
Q: Community cloud--will you have per hour pricing.
A: We need for this to be competitively priced so I think you'll find that. We need use cases for that.
Q: Suppose we have a service that we think would be useful to the community?
A: Email Jerry. We would like to have a fully worked out process for submitting and vetting new ideas that are scalable and broadly interesting. We'll get there! We want to be open to experimentation.
Q: When will all the pricing and information be available?
A: Go to the I2 Net+ page. That is being updated as we go along. You will be able to brand the Box service so that it does not appear that you are promoting Box.
- Web Site
- Dave Lambert, CEO of Internet2
- Joe Feddoso, MCNC
- IDEA Awards Recognizing Exemplary Applications that Use Advanced Networking
- Internet2 Strategic Focus Area Update
- Services Above the Network
- Research Collaboration
- Network Update
- Host: MCNC
- New Developments with Internet2
Thanks to program committee, sponsors, and welcome to new members.
MCNC has changed: Instead of being the exclusive R&E network for higher ed, MCNC is becoming the network of a broader scope of organizations. The percent of traffic over the past five years has gone from being predominantly from one or two institutions to many more.
Think about collaboration across the network, collaboration to build the network and collaboration enabled by the network.
Tom Knab stepped up to present the four IDEA awards.
- Magnitude of + Impact
- Breadth of Impact
- Technical Merit
A collaborative Network Fault Diagnosis System by Kyung-Hwa Kim, Ph.D. student at Columbia University.
Monitoring and Visualization of Energy Consumed by Networks by Baris Aksanli, Ph.D. student at University of California at San Diego.
A Telematic Opera by Scott Deal at Indiana University-Purdue University and Mathtew Burner, University of Virginia.
On-Demand Secure Circuits & Advance Reservation System developed by a team at ESnet with partners at I2 and DANTE.
Presented by Jerry Grochow and Shelton Waggener.
Previous developments in this area
- Internet2 Commons
- Cloud Services Platform
Collaboration has lead us to a partnership with box.net. It's a service where we will ultimately have
full ability to use campus authentication to access this service. We heard from the COO of box.net who showed a video on how box works.
This will allow campuses to subscribe to the box service at an incredible savings and it will be opened to the community in 2012 to members of both I2 and InCommon.
This will be tested with Indiana, Penn State, Notre Dame and other universities with both research and administrative computing functions.
Meg Whitman, President & CEO of HP, did a live feed from California to kick off this effort between HP, SHI, and Internet2.
We heard from others, as well.
Presented by Steve Wolfe: New members of the staff include Wendy Huntoon as Chief Architect, Scott Brim and Jason Zurakawski.
Vision is for I2 to be considered a vital part of their research project. Want other networks to be considered an important partner.
Strategy is to create an environment of collaboration; make I2 more user friendly; convene, confer, consult and participate as part of the research community; offer visible, proactive, comprehensive researcher support from proposal to publication; translate participation into new see ices that can be incubated, tested and deployed; (one other bullet point that was skipped).
There is a strategy for chief technology office at Internet2 to provide researcher support. There will also be a rotator program where someone from a campus, RON or agency can work for Internet2 for a period of time and then return to his original location.
ESnet is reaching it's 25th anniversary. There was a $62M ARRA award to ESnet.
A 100g footprint will be built out in 2012.
Robb Vietzke: First 100G transcontinental network (went live ahead of schedule?) There will be a joint Internet2/ESnet 100G transcontintental network.
Deeper collaboration with GENI is planned.
Q: What about security for the net+ services?
A: I2 has a team that is knowledgeable about security. The providers also spoke to the security of their systems.
Q: The overlap between the community cloud and the private cloud that many of us are engaged in may require some education, for example, what is the benefit of this rather than what a researcher may already be using?
A: We found that there's a language issue about what is a cloud and how do we use it. The broader community may be looking to the education community to set standards in some ways.
Q: From the perspective of CTO, what is your posture towards scholarship?
A: It's critical that as we move this forward, it's a scholarship platform, not just a research platform.
Reasons for not adopting broadband
- no access
- can't afford
- can't use
- don't want
Goal is to address the last two.
Set up computer centers throughout the State of Michigan, mostly libraries, over 200 in all.
- Deployed over 2000 computers (2/3 of target)
- over 15,000 people trained
700 square miles
6000 work stations
How to provide centralized service to all that is fair and uniform?
- IBM provided a virtual desktop as a service solution.
- Single pooled student image delivered as Enhanced Internet Access (EIA)
- Cost effective
- Familiarity with E-rate funding
- ability to co-locate infrastructure near customer premises
- ability to add more work stations at a flat rate
- improved desktop performance
- equalized student access
- home access to teachers or home bound students
- enhanced services (web clients, I-pad, Android, Wyse thin clients)
Started with vmware/desktone broker
Now, Citrix/Xendesktop broker
Did a number of things to connect to University of Minnesota and I2 as early as 2001.
- Started doing things via Internet with remote universities. Have their own hub room.
- In 2006 received an award to begin partnering with hospitals.
- Now connect to a network of Minnesota universities, public agencies.
- Instruction - partnering with high schools
- sharing of coursework
- expanded access to resources
- science lab
- Resource sharing - e.g., virtual servers
- Training of incumbent workforce in healthcare and manufacturing
- Industry partners access Pine Tech courses
- Parnerships come first
- common need
- shared vision
- enlightened self interest
CCs are backbone of economic development in North Caroline.
- 3rd largest in US
- 1/8 of citizens attended a CC
- NC Law: No one can live more than a 30 minute drive from a CC.
- NC colleges, because of its size, provides a lot of data to researchers and funding agencies studying/assessing success.
We have to respond to all education needs across the state.
Primary business driver for schools is around learning.
MCNC has been a wonderful partner as part of a rural broadband initiative.
Two BTOP awards have expanded the network throughout the state of North Carolina.
- Virtual Learning Community that encompasses course development, professional development, course delivery
- Faculty get access to enhanced resources
- Educate legislators -- biggest challenge -- explain why CCs need to be part of the Internet2 and state R&E network.
- Coordinate with college campus academic schedules
Created with state funding 10 years ago with 150 MB 10 years ago.
Stretching that capacity, now.
BTOP projects are going in to raise capacity.
Tied into LEARN (Texas R&E network)
Provide 25 to 75 Mb connectivity to network
- Do a lot of remote/online course work
- Do a lot of Video Conference bridging
- Continued growth by providing college courses to high school students (dual credit courses)
- This has increased the number of high school students who go on to college and complete a 2-year degree, at least.
- Nursing education is vital and it's being sent to high schools, hospitals
- Video classrooms -- connecting over 300 classrooms
- Need continues to outstrip capacity
- Collaboration and shared control of processes are greatest challenges
- External funding makes the case for doing this--when the state or feds offer funding for something like this, then the folks within the state see that it must be important.
USUCAN report by Mark Johnson, Interim Director of USUCAN project for Internet2, and Mike Roberts, Chair of the Task Force on Community Anchor Network Economic Models
- This was a great time to get funding. There have been questions about wisdom of this, but it's now or never and then build on what we have
- Consolidation of USUCAN with FCC Rural Health within Internet2
- USUCAN affiliate program being discussed--pilot will follow
(Note: Some answers were provided by others than Mark.)
Q: How can R&E's help you with outreach and when will we have a better sense of when we can get started with you. We're not sure how.
A: We need help in defining this affiliate program.
Q: How does this relate to SEGPs?
A: We roll the SEGP program into this and hopefully expand the program. It will have dedicated staff and won't be mixed with the research part. We want to understand the costs and then make decisions regarding costs, later. Unit costs should hopefully go down.
Q: What about advocates and the national office indicated in the report?
A: Not our intent to be a lobbying organization. We will be informational--providing information to people in DC.
Q: How will organizations be related to each other and will there be independent governance for USUCAN?
A: We're starting with a council. Now we're going to have eight types of organizations involved, and the issue of governance is critical. Governance of I2 is a bit top-heavy--we need some governance light. As I2 incubates things that are outside of the core mission, we need to have a way to handle this.
Q: Is there anything we in the K20 group can do during this meeting to help you out?
A: K20 could help to assure that we're representing others by giving us some names and attributes of what a successful council members would look like as well as notes on process for identifying them. Feedback on how costs should evolve would be welcome.
During the Spring 2011 Member Meeting we held a session discussing the process through which the Great Plains Network made broad support for regional cyberinfrastructure a strategic priority, and how more generally RONs provide a base for developing CI support organizations. Based on the positive response to that session we propose a follow-up discussion in which several regional CI groups associated with their RONs will compare organizational models and operational details. In the past, much of the focus on expanding cyberinfrastructure abilities and capabilities has been at the campus and national levels. Lately, there has been a trend for existing regional organizations (e.g., the Great Plains Network, SURA) to extend support to the development of regional cyberinfrastructure as a strategic priority. This panel focuses on the role of these intermediate-level organizations -- RONs, state and other regional organizations -- in advancing priorities at the campus and national levels. Participants include representatives from the Great Plains Network, SURA, RENCI as well as campus representatives who will discuss organizational models and operational details.
- Stan Ahalt , RENCI
- Paul Avery , University of Florida [pdf]
- Gary Crane , SURA
- Rich Knepper , Indiana University
- Rick McMullen , University of Kansas
- Greg Monaco , Great Plains Network
There was a great deal of interest in the topic as measured by the increase in audience from the Spring.
Some of the more interesting questions had to do with collaboration across wider distances vs state-wide or even more local (i.e., RENCI). The concensus was that face to face opportunities to get together and get to know one another are very useful to help individuals to work successfully together.