Information Technology Planning Board (ITPB)
Meeting Summary
Tuesday, Jan. 29, 2008
www.itpb.ucla.edu

ITPB Attendees:
Kathryn Atchison, Christine Borgman, Russel Caflisch, Alfonso Cardenas, Mitch Creem, Jim Davis, Michael Goldstein, Xue Hua, Bill Jepson, David Kaplan, Kathleen Komar, Sam Morabito, Warren Mori, Marilyn Raphael, Alan Robinson

Guests:
Ross Bollens, Carmela Cunningham, Larry Loeher, Sean Pine, Nick Reddingius, Margo Reveil, Terry Ryan, Mike Schilling, Libbie Stephenson, Kent Wada, Don Worth

Agenda item 1:  December 14 2007 Meeting Summary

The December 14, 2007 meeting summary was approved as presented.

Agenda Item 2:  UCOP Planning Update (A.  Report on Cyberinfrastructure Implementation Report and B.  Long-Range Planning Document)

A.  Jim Davis reported on the UCOP ITGC Report and specifically on the Cyberinfrastructure Implementation recommendations, which focused on three main issues:  1) Models for UC-shared cycles, storage, grids and access to applications, 2) Analysis of the network layer of IT infrastructure at campuses, and 3) Data Centers.  While the report discussed information technology for research, education, collaboration and administration recommendations on research CI and the associated infrastructure were the most defined. 

Report Premises:

• The cyberinfrastructure is critical to all researchers, which includes a handful of “super users” across the UC system who are doing petascale computing, are using the national labs, and are on the  “bleeding edge;” CI-enabled, broad-based users who are doing both “hard” and “soft” research; and researchers who are working on the cyberinfrastucture.  One question is how does UC leverage resources for the super users.
• While the national labs are co-located with UC – which gives us special advantage – they are not UC labs and our researchers are among a national set of users for shared access.  The San Diego Super Computer Center involves shared access for researchers across the country.
• Research trends show that NSF, NII, DOE are all moving toward large-group research, which necessitates virtual research groups (60 – 80 person teams).  This creates the requirement for a collaboration piece that can accommodate applications, storage, and cycles.
• The UC grid will benefit both capacity and capability.
• It needs to be validated but we believe there is unused computational capacity across the system.  We have about 30 – 40 percent utilization, and we need to look at increasing utilization.  This is happening at the same time that the data centers are tapped out.
• It is not possible to fund the computational resources needed out of a single resource grant; by aggregating resources we can build larger capacity resources than those purchased for a single grant.
• The premise is that research benefits from ready access such as is available through the grid. 

Discussion Points:

Data Centers

• There’s a big energy cost incurred when individual users put clusters in closets, rather than in a data centers.  Compute nodes being housed outside of data centers are not efficient.  Unused compute nodes still draw 80 percent power.  If the nodes are housed in a data center and part of the grid, they could be used more efficiently.  Individual researchers also have to pay people to manage their clusters, which is a cost that could be mitigated if their clusters are in a data center, ATS let’s researchers “buy in” and share cycles, which is a big benefit.

• There’s a need to look at constraints on grant-purchased machines; there might be issues about appropriating unused machines purchased from one grant on another grant.  Is the “shared cluster” model acceptable to granting agencies?  Federal grants specify that machines belong to the university, and collaboration is in line with the NSF model.  NSF does have concerns about funding things that are wasted; which should mean that sharing clusters would be appropriate.  The Board agreed there should be follow-up on this question.

• One question about data centers is the need for an “end of life” equipment strategy.  As an example, new technologies have been identified that are four times more energy-efficient than 18-month-old technologies.  What’s the policy for having old, inefficient technologies moved out of data centers?  The Math Science data center specifies a three-year lifecycle and then requires an assessment every year to indicate if machines should be retired.  If someone has new money to buy better equipment, the oldest, least used equipment goes.  This is a good reason to buy fewer machines and replace them more frequently.

• New units are not easy to program.  If researchers work with experts (i.e. ATS High Performance Computing experts), they’re able to optimize their code, which means they need to buy less equipment to run their jobs. This is another way to increase utilization.

• The Question is how to build out in the best possible way, It’s not easy to get funding for infrastructure.  A facility and major equipment may get funded, but there’s also cabling, racks, etc., which tend to go unfunded in grants.  Getting funding agencies to start funding these kinds of things is a long-term process, but it has to be started.

• A related discussion on overhead rates included the question of whether granting agencies would fund time for technologists and the unfunded necessities that are required in a data center.  The point was made that the UC overhead rate has always been lower than the campus actually spends to support researchers, but that the federal government resists raising overhead rates.  It was also strongly noted  that increasing the overhead rate would make UC researchers significantly  less competitive for future grants.

The Grid

• The CI Implementation report recommends developing UC Grid Research Cyberinfrastructure services. The model is for a “dynamic grid,” in which the grid is provided by the institution, with resources on the grid are owned by specific research groups.   The goal is to encourage and support autonomy.  Groups of experts can build more collective value for individual researchers.  UCLA is leading the UC grid effort which now has five campuses on its grid.  Key discussion needs to focus around models and policies for future grid development.

The Network

The report noted that the UC network infrastructure must be brought up to speed.  The network as a whole is in great shape statewide with connectivity among campuses.  However, there are bottlenecks within the institutions.  UCLA is at 10Gig connectivity but others are  not.   $70 million is the cost estimate to bring all campuses up to speed.  Discussion revolved around how the funding should be collected and allocated for bringing individual UC campuses’ network infrastructure up to speed.  It was noted that some campuses – such as UCLA – have enhanced their networks at their own cost, rather than out of a common pool, and that funding other campuses out of a common pool would not be equitable.  It is playing out that way, that campuses are paying for their own network infrastructure upgrade.  It was also brought up that some campuses don’t have the resources in the first place to pay for their own network upgrade.  System funds should be may be used to bring connectivity to campus but upgrades within campus should be self-funded.

It was agreed that a long-range view of the cyberinfrastructure is needed and we need to look forward to anticipate where we will be in 2025.

B.  Jim Davis reported on the UCOP long-range planning report.  This report also addressed data centers, networks and the UC Cyberinfrastructure, and Davis noted consistency between the two documents.

Agenda Item 3:  Report from Research & Educational Data Management Working Group

Christine Borgman, Chair of the ITPB Subgroup on Research & Educational Data management, presented a summary of the discussions in the subgroup’s meetings.

A data management infrastructure can provide help for researchers to:

• create data management plans
• develop institutional responses to grant agencies regarding data
• create policy and standards for setting up databases
• identify who owns the problem

The committee recommended developing an institutional response to the directions of grant agencies regarding data.  One recommendation involved hiring a GSR to help the committee assemble a survey, compile results and recommendations, construct and maintain a website and provide coordination of education and referral services.  Discussion followed about how ITPB should go about securing funding for committee recommendations.

A motion was made to recommend hiring a qualified GSR possibly by the CIO’s office to assist the subcommittee on this set of goals.

Discussion ensued on whether ITPB has funding to hire staff and whether the Board could unilaterally assign funding.  It was suggested that the request be tabled until all working groups could put forth their requests for resources.

The motion was made that all working groups should come back to the ITPB with their funding requests and that the ITPB should put forth a budget that would include all ITPB-approved requests and take it to the Academic Senate and EVC Scott Waugh. The motion was seconded and passed unanimously. 

Agenda Item 4:  Report from the ITPB Wireless Working Group

Jim Davis gave an update on the Wireless Group Meeting, which focused on three points:

• Beginning a discussion on planning wireless in a holistic matter, including how to build wireless in a “mobile” way that truly addresses mobility, as opposed to addressing wireless in a “cordless” way.

• Understanding what the city of Los Angeles is doing about wireless and tying into that plan

• Addressing the student survey on wireless.

Wireless working group member Mike Schilling requested approval of the ITPB to use existing TIF funds to do an assessment that would inform a request for a cost estimate on putting wireless in the areas identified in the student survey.  Jim Davis motioned for ITPB approval.

Discussion explored the funding authority and capacity of the ITPB, and the following points were made:

• There are other committees that deal with IT fund requests – such as CITI.

• The specific wireless committee request was not for CITI or TIER funds, but rather the request was to use available CTS TIF funds for the assessment. The difference between the Data Management Working Group request for funds and the Wireless group request is that the Data Management Working group was asking for funds, while the Wireless group was asking for “approval” on using existing funds belonging to a central organization as a one-time service expansion.  The request was really to “vet” the usage of TIF funds for the assessment through the ITPB.

• It was pointed out that the ITPB should positively respond to the student survey on wireless. 

• The discussion then turned to the role of the ITPB, which was called a policy, planning board – not a funding board - and how to manage the funding requests of the other working groups.

• It was also pointed out that the ITPB has agreed to review TIF service changes. This particular request was treated as new one-time service request.

• Chair Alfonso Cardenas reminded the board that the ITPB charter also includes advising on IT investments.

• It was stated that the ITPB must “pay attention to the subject of funding, or all the work of the subcommittees will be for naught.”  The suggestion was made that the ITPB should assess the recommendations set forth by the subcommittees and prioritize and legitimize those recommendations.

• The motion was made to use existing TIF funds to do an assessment – with no increase in the TIF – to get a cost estimate on what we can do in the short run to put wireless in these areas that the students requested.

Agenda Item 5:  Reintroduction of Requirement for Faculty and Staff to have Email Addresses on File and in the Directory

Kent Wada introduced the document on Use of Email for Official Communication with Employees and Students. In the past the ITPB had suggested that UCLA move to use email as the primary method for official communications with employees and students.  The ITPB had previously endorsed the concept that faculty and staff and emeriti faculty have permanent forwarding of their email. 

Kent Wada also introduced the Email Address on File document pointing out that in order for it to become a reality, that:  1) all employees and students would have to have email addresses on file and 2) there was a need to gather all email addresses and keep them in a separate internal database (which would not be available to the public, would be unsearchable, and not a directory).  There is a problem with using the UCLA directory to provide emails, as not everyone is listed in the directory and some people have good personal reasons for not being listed. 

Action:  Kent Wada will lead effort to develop a memo to staff and faculty that will be distributed over EVC Scott Waugh’s signature.

The meeting ended at 11:05 a.m.