Duke ITAC - March 4, 2010 Minutes

Duke ITAC - March 4, 2010 Minutes

ITAC Meeting Minutes
March 4, 2010, 4:00-5:30
Allen Building Board Room
  • Announcements
  • Computer Lab Directions (Samantha Earp, Tracy Futhey)
  • Lecture Capture Directions and Experiences (Samantha Earp, Ed Gomes, Wayne Miller, Andrew Schretter)
  • Google Fiber Initiative (Julian Lombardi, Kevin Davis)


Announcements

Terry Oas called the meeting to order with an announcement about campus outages over the past two weeks. Representatives spoke on behalf of both ePrint and dCal regarding recent outages, and Carl McMillon spoke to a 3-4 hour power outage experienced in two non-enterprise data centers impacting departmental servers and the Duke Shared Cluster Resources was attributed to a blown fuse in a high-voltage transformer in the North Building.

Minutes from the February 18th meeting will be distributed with the minutes from this meeting for approval at the next ITAC meeting.



Computer Lab Directions (Samantha Earp, Tracy Futhey)

Tracy began by thanking special guests and lab users Mike Gustafson and Dalene Stangl for coming to participate in this discussion.

As Tracy explained, most intra-departmental efforts to cut costs have already been addressed, and the Duke Administrative Reform Team (DART) is now turning its efforts toward saving money in IT-related functions across Duke.  Currently, the team aims to estimate how much the university is spending on labs and whether alternative approaches exist that could reduce these costs now that nearly every student comes to Duke with a personal laptop.

The group, which included student and faculty representation as well as ITAC-appointed IT representatives, has surveyed the layout and usage of public computing devices currently offered at Duke, including kiosks, general purpose machines, classroom labs, virtual labs, and specialized-purpose labs.  In total, there are an estimated 1,700 computers installed between physical and virtual labs across the entire institution, accounting for $1.6 million in operational costs per year.

Current reasons for use of computer labs include convenience, group study, classroom activities, and access to specialized hardware or software.   Historically, Tracy noted, physical labs have been offered in part in order to put students unable to purchase a computer on equal footing with those who have computers of their own.  These days, however, estimates are that fewer than 4% of students arrive on campus without a computer, and only about 5% of students with personal computers have non-mobile computers.  With an overwhelming majority of students carrying laptops, we may not need the number of computer labs currently being offered.  DART’s research supports this, Tracy pointed out; OIT kiosks are used less than one hour per day on average, for instance.

Applications most utilized in Duke computer labs are general productivity software (e.g., browsers, email, word processing).  Linux computers often see usage in math and statistical packages, but programming, multimedia, and GIS/visualization packages were launched only occasionally.  

DART’s recommendations, as recounted by Tracy, included encouraging laptop purchase among undergraduates; virtual machine (VM) access to replace select underutilized labs; fewer public computers and kiosks in Perkins/Bostock, general purpose labs, and residence hall facilities; promotion of existing OIT classroom labs, enhanced support for limited OIT specialty labs, and reduction of non-ePrint public printing and general purpose and teaching computers in schools and non-OIT labs.

Estimated savings for implementing these changes could be as much as $800,000 in operating costs per year if 90% of lab desktops were replaced with virtual machines on a one-to-one basis.  Additionally, eliminating free printing for non-Duke patrons could save an estimated $100,000 per year.  Total savings across all the proposed initiatives could account for over $1M in savings each year, based on modest demand lab/VM reduction assumptions, and after taking into account that new programs may be required to support these changes, such as a laptop lending program for students who find themselves without a usable machine.

Tracy explained that the team intends to evaluate these changes a step at a time in order to avoid discontinuity of service.  For the time being, an ITAC working group comprised of faculty, students, and administrators will be meeting in order to review lab utilization, assess proposed changes in labs, and review VM pilot results.  A working group will be convened in each of the next three years to evaluate direction and adapt as may be necessary.

As for OIT labs, current plans include a virtual machine pilot, moving the Multimedia Project Studio from Old Chem to the Bostock Technology Alcove, enhancing standards for equipment in key labs, and heavy evaluation of lab usage and coordination/scheduling of classroom labs in particular.

Tracy then opened the floor to questions.  Alvy Lebeck asked about the usage model for the proposed VM switch.  Tracy responded that the assumption is a one-to-one replacement of a mid-level VM per phased-out desktop.  The need for VMs may not be quite this great, so the one-to-one guide can be considered a starting point, she said. This also assumes equal software cost between the two machines, so the potential savings could be greater depending on licensing and need.

Dalene Stangl expressed concern about the technical challenges of a virtualized classroom lab, citing difficulties with keeping a group of laptop users on the same page with as few as ten students.  Michael Gustafson shared a similar concern about added startup time in classes as a result of these changes, as well as a burden that may fall on faculty to provide technical support to students struggling to access a VM on a personal laptop.  

Samantha Earp clarified that these figures don’t assume elimination of all physical specialized machines, and that teaching labs will still be offered where appropriate.  Among other concerns that an immediate switchover to virtualized labs and classrooms could become chaotic, Tracy reiterated that these initiatives will not be implemented all at once, and a phased approach is intended to ensure a smooth transition between our current model and the proposed changes.  Also noted was that NC State has a model very similar to the proposed adjustments and has had great success with it.




Lecture Capture and Experiences (Samantha Earp, Ed Gomes, Wayne Miller, Andrew Schretter)

Samantha began with an overview of the services offered by our current lecture capture solution, Lectopia.  As Lectopia’s vendor is phasing out the software, a new product has been selected to replace it.  In the four years that Lectopia has been in use at Duke, lecture capture has gained quite a bit of momentum and new needs have emerged as a result.  Popular feature requests include live streaming, portability, usage tracking, lightweight editing, and easier connections to other digital media systems.  

As Samantha explained, the evaluation team recently completed phase 2A, wherein Panopto was selected as the Lectopia’s replacement from an original pool of 8 products.  Of the finalists, Panopto most impressed the team on the basis of a responsive support model, interactive note taking feature, live streaming, capability for high-quality capture and multiple recording sessions, and a personal capture piece, which can be installed and run on any personal computer.
Now that a successor to Lectopia has been selected, the team is working toward deployment, developing documentation and a service model for Panopto’s software component, and organizing training and file migration from Lectopia.  Other identified tasks include development of a data retention policy, mechanisms for sharing content (where authorized and appropriate), more coordination with the CIT on best practices for classroom implementations, and coordinated planning on classroom A/V integration.

Samantha then showed results from a survey sent to faculty to gain insight into how instructors at Duke see lecture capture and what they would like to see from a lecture capture program. She then opened the floor for questioning.  Both Samantha and Steve Toback responded to questions about Panopto, confirming that publishing lecture media recorded via a personal laptop will be possible with the new product, and discussing the possibility of recording computer output, such as notes written on a tablet computer.  Responding to a question from DSG representative Michael Ansel, Steve explained that the Panopto software can capture both presentation slides and chalkboard notes at the same time, but that the recording is limited to a maximum of two video streams.

Ed Gomes and Wayne Miller then discussed their experiences with Lectopia, which they agreed had worked well for them, although Wayne noted the Law School was also able to record lectures prior to DukeCapture thanks to a control room connecting 11 of their 12 classrooms.  Ed and Wayne discussed logistical challenges, such as having to install spotlights in order to be seen properly on camera, and noted a need to be able to record students presenting material.  And while many faculty have been resistant to the idea of digital lecture capture, Ed explained that after being asked to record their lectures during the H1N1 pandemic, faculty who previously did not see the need for lecture capture are now big supporters.  Wayne added that Lectopia’s successor needs a self-service option in order to make these technologies more accessible to faculty.

Andrew Schretter then described a Flash-based lecture capture system implemented by the Math department that uses a pan-tilt camera and pulls from many different sources. To address the often-voiced concern about capturing everything going on in the room, Andrew plans to include a high-resolution digital camera that takes a picture of the room every 30 seconds in order to provide a high-level view of what’s going on in the room at any given time in the lecture.

One benefit to Andrew’s system is that it’s very inexpensive, there is a high degree of control over the content, and it’s extensible with in-house development.  This system can be viewed at https://www.math.duke.edu/video/video.html, and more information about the digital camera capture implementation can be found at http://www.openeya.org/.  In response to a question from Terry Oas regarding viability of this solution for use across Duke, Andrew indicated that while his system is trivial to expand on a smaller level, it may not be scalable at the enterprise level.

Steve mentioned that ad-hoc scheduling of lectures is a weakness of Lectopia.  Andrew’s system records a given room all day long and then splits the recordings into session-specific files.  Steve added that Panopto handles ad-hoc scheduling much better than Lectopia and a site license will be available for all faculty, staff, and students.



Google Fiber Initiative (Julian Lombardi, Kevin Davis)

Julian Lombardi introduced Google Fiber as an experimental project intended to deploy a fiber network to a trial location within the United States.  Google’s interest in this initiative is better understanding the logistics of such fiber deploys and more importantly, what kinds of web applications would be possible in a world where Internet connectivity is ubiquitous and strong.  The proposed network’s capability for speed – a gigabit per second – is about 100 times faster than most home broadband connections.

The benefits to being among the first communities to have a fiber network go far beyond web browsing experience.  Durham strikes many as a great choice for this program because of its relationship to RTP and universities, the technical and entrepreneurial nature of the city, and a marked digital divide within the community.  A strong common network would give Durham an even stronger technological presence as well as give all residents affordable access to take advantage of services developed as a result of this new technology.

While the benefits to a local fiber network are great, so is the cost.  Without Google’s help, such a fiber network deploy would cost Durham County an estimated $600M.  Kevin Davis provided a link for the HiFiber Durham Initiative webpage (http://www.hifiberdurham.com/).  Durham’s public response to Google’s call for interest, HiFiber Durham hopes to demonstrate to Google that Durham is the best candidate for the Google Fiber program.

Julian and Kevin attended a recent Chamber of Commerce meeting to discuss action being taken to attract Google’s attention toward Durham.  Google has sent out a request for information from city and state governments regarding their willingness to participate and the physical layout of their cities in order to be considered for this deployment.  Durham’s bid for selection will involve a three-pronged approach: providing Google with extensive information about the county’s physical and technological layout, coalescing community response and show that this program would be welcomed by the community, and providing details as to the kinds of innovation this community could bring forward if given access to such a network. Kevin Davis has been leading efforts to compile ideas for new technological innovation possibilities, and encouraged the group to submit their ideas via the HiFiber Durham website.

The deadline for submission of information to Google is March 26th, and Google reserves the right to select a community earlier, so time to make this case is limited.  All ITAC members were encouraged to visit the HiFiber Durham website to nominate Durham for this program and submit ideas.  Julian encouraged the group to think critically about how a fiber network could change Durham and how we could do things differently were one brought to our community.