Duke ITAC - April 9, 2009 Minutes

Duke ITAC - April 9, 2009 Minutes

ITAC Agenda
April 9, 2009 4:00-5:30
RENCI Engagement Center

  • Announcements & Meeting Minutes
  • Perspectives on IT - Dr. Peter Lange
  • Introducing Paul Horner, IT Security Officer
  • OIT Metrics Initiative (Lew Kellogg, Susan Lynge)
  • Changes to Port 25/SMTP Configuration (Klara Jelinkova)
  • New Webmail Directions (Mark McCahill)

Announcements & Meeting Minutes

Terry Oas opened by asking ITAC members present at the March 26, 2009 meeting if they had comments on the minutes. Noting no objections, Terry accepted the minutes and stated that they would be posted on the ITAC web site.

Perspectives on IT - Dr. Peter Lange


Terry introduced Provost Peter Lange.  Terry mentioned Provost Lange’s presence was part of a continuous effort to get feedback about IT issues from both the Provost’s Office and the Office of the Executive Vice President Tallman Trask. 


Dr. Lange referenced a set of questions he was provided in advance and invited discussion based on his responses to those questions or other topics.


The first question was around the Duke Administrative Reform Team (DART) and its potential impact on IT.  In addition, how could IT contribute to cost reductions elsewhere?  Dr. Lange stated that IT would be an important component in DART’s efforts. He commended Tracy for being proactive in identifying areas DART could look.  He added that he did not anticipate adding to OIT’s current efforts to identify cost reductions.  Dr. Lange mentioned that issues in this space frequently revolve around centralization and de-centralization.  He reminded the ITAC members that he previously presented to them on being “a de-centralizer”; nevertheless, he noted that efficiency, security and cost tend to be the main benefits of centralization, with cost being the third most important.  Some services lend themselves better to centralization, for example, e-mail, he said.  He suggested that this would likely lead to efficiency and security improvements, but he said overall cost savings may be harder to capture and represent for individual units.


As for how IT could contribute to cost reductions elsewhere, Dr. Lange said that he has yet to experience that.  He described as “revolutionary” the possibility of IT reducing institutional costs globally and being able to capture those cost reductions.  He added that specific cost reductions are frequently achieved; however, they are usually accompanied by a larger cost increase, thus resulting in a net cost increase.  Notably, these efforts can lead to educational or research advantages, but IT innovation has not traditionally been a source of cost reductions.  Dr. Lange added that the DART process is not solely a cost reduction effort. The DART process aims to improve overall service delivery in terms of Duke’s global mission, he said.


Terry said an example of a current cost would be the amount of time administrative assistants spend scheduling meetings.  He suggested that broader use of the central calendaring system could help reduce cost.  Dr. Lange said that his office schedules many meetings, and he suggested that asynchronous meeting scheduling, independent of software solutions, is ineffective.  Dr. Lange’s scheduling staff has begun having conference calls between all admin staff for a given meeting.  He described this process as “vastly faster.” Dr. Lange added that this might not necessarily reduce costs, but it saves a great deal of time the admin staff can then use for more important efforts.


Dr. Lange moved on to an update around Duke University’s internationalization efforts and its impact on technology.  Dr. Lange stated that overall goals and locations remain the same, but that the timeline may be extended. One advantage is our partners’ investment in facilities costs, Dr. Lange added.  The IT demands have not changed, but the internationalization task force is focusing more intently on those demands, he said.  For example, the task force is building more into the development proposals for the various sites.  He added that Duke may soon engage a firm for operational assistance.


John Board asked about the visions for how technology would support the international educational mission.  For example, one vision would be to have all campuses simultaneously connected via HD for real time multi-site classes.  This is an expensive vision, John said, and asked who would be articulating that vision and when.  Dr. Lange said the international initiative leadership would define that.  Dr. Lange said that the vision is similar to what John described.  He added an example would be Duke’s “Global Semester Abroad” initiative.  Students in Delhi and Shanghai could travel to the other locations for six weeks and then switch.  However, the experience would be richer if students could communicate with students at other sites while traveling.  Dr. Lange suggested the ability to deliver this depends on who Duke partners with.  He noted that scaling a spoke and hub solution to a global level would be difficult, and he is unsure if it could be achieved by 2010, but it is an exciting vision.  Dr. Lange noted that we are expressing a global vision that needs an infrastructure that is commensurate with it.


Dr. Lange reviewed the Link assessment report and stated that he viewed the Link as a success.  He commended everyone who contributed to its delivery.  He described the Link as a more successful learning environment than as a technologically enhanced learning space.  The main issue, he noted, was the reliability of the point-to-point joint courses using online conferencing technology.  His expectation was that over the coming years he anticipates more joint educational efforts between Carolina and Duke.   Dr. Lange said the fundamental goal is to have technology that is intuitive enough that can be used by faculty that are not technically savvy. 


Another comment from the Link report that Dr. Lange highlighted was the challenge for students to be able to synchronously display items from their computers into a shared space. He said the environment is great and is always praised; however, the next goal should be to bring the Link up to our technological aspirations.  The goal would be to have the Link be our model site that other sites want to emulate. 


Terry asked if Dr. Lange envisioned the Link as the site for supporting high-end video conferencing related to our international efforts.  Dr. Lange responded the Link was one site configured with these capabilities; RENCI and Fuqua were others.  Furthermore, he anticipated that Duke will need more.  He said that our physical and technological capabilities should not inhibit our creativity.  Terry asked if someone could summarize the technological bottlenecks.  Ed Gomes said it was the consistency of the technology.  Tracy added that everything is proprietary and that raises compatibility issues.  Robert Wolpert asked if the envisioned future would be dedicated rooms or students anywhere with their laptops.  Dr. Lange said it was likely a mix.  He added that ten students with ten webcam-enabled laptops is a different experience from ten students sitting together in a single room.


Dr. Lange turned to the last question regarding thoughts on in-house versus outsourced web design. He said a web designer rich community such as the Research Triangle area, Duke could train and “license” these groups to provide demand-responsive web design services. He noted that Blackwell frequently reaches capacity.  It is expensive to add FTEs for a service with highly fluctuating demands with base standards for which others could be trained, he added.  He added that the current fiscal environment might lead Duke to this as a possible model.  In effect, he foresees the possibility of internally organizing and managing the process of standards, training, and vendor relationships.  Duke’s decentralized web design lends itself well to this different model, he said.


Wayne Miller stated that the consistency of a development style, the availability of skilled staff knowledgeable in the specific environment, and in house programming support are crucial to the maintenance and support of the school web site.  He added that this in house knowledge adds efficiency to the school, and ultimately to the University.  Dr. Lange said many organizations have shown that an in-house and outsourced solution mix can better meet the capacity fluctuations at a lower cost.  He added that the larger pool of designers ostensibly would increase creativity. 


An ITAC member stated a concern about the assumption of fluctuating demand, and postulated that demand is growing steadily despite near-term fluctuation. He asked when Duke might have sufficient demand to justify the full-time overhead and how to identify that.  Dr. Lange stated Duke’s overall web design demands were unclear.  He said that this ties in to one of DART’s goals.  Dr. Lange added global demand might be increasing, whereas local demand likely fluctuates.


An ITAC member agreed with Dr. Lange’s assertion that groups tend to staff to peak demand, nothing that the School of Nursing decided to staff to the average demand and call on Blackwell for peak needs.  The web team is a combination of web developers and project managers.  Another advantage of this model was the ability to call upon specializations as needed rather than attempting to meet all technical needs in house, he said. 


Dave Richardson questioned the basis of the idea of fluctuation in demand.  He suggested web sites follow a “build, cruise, decay, crash, and rebuild” model.  That is, sites are built but fail to be maintained for a variety of reasons and then are retired.  He said Dr. Lange’s model could reduce the instances of “rebuilds” and encourage better ongoing maintenance.  Dr. Lange asked the group the maintenance demands of the technology versus the content.  Tracy said content takes the most continuous effort and design refreshes are periodic.  Dave said it would be beneficial to have a central location to mediate the balance between content and design needs.  Terry asked if Blackwell did this.  Tracy confirmed that they do.



Introducing Paul Horner, IT Security Officer


Tracy introduced Paul Horner.  Paul recently presented to ITAC on the security scanning initiative (http://www.oit.duke.edu/itac/minutes/2009/90226minutes.html).  Paul has joined the CIO office (not OIT) as the new IT security officer for the campus.  Paul will lead an effort to appropriately focus Duke’s security efforts. 


Tracy asked Paul to share any insights from his first weeks in his new role.  Paul said he has experienced a broad range of security issues, that he suspects is just a coincidence, and has dealt with both tactical and strategic matters.  He added his pleasure at inheriting a great team and being able to collaborate with Klara Jelinkova who had served in an interim capacity.  Tracy also thanks Klara for her time in the interim role.



OIT Metrics Initiative - Lew Kellogg, Susan Lynge


Lew Kellogg said OIT is working on a metrics initiative with a goal of maximizing IT’s value to Duke.  Lew said OIT wanted to collect the appropriate information to help make intelligent decisions.  In addition, the initiative would enable OIT to proactively identify capacity and demand constraints. 


The metrics team defined four factors to consider: value, performance, quality, and compliance, he said.  The effort is looking at both qualitative and quantitative measures. The team divided their efforts into three areas:

  • Operational – Systems Monitoring
  • Tactical – Trend Analysis
  • Strategic – Long Term View

Lew said the team identified three pilot services: email, paging, and networking.  Collaborating with other groups, the metrics team began to ask “How do we know we know that this service is doing what it is supposed to do?”  Based on these three pilot efforts, the team collected information and experimented with different presentation methods.  Lew said the team is ready to operationalize the monitoring of these metrics to help detect trends.  Lew added that in the Networking area, OIT partnered with Pratt to develop a model to predict DHCP failures.  Finally, to provide a strategic view, the metrics team is developing a metrics dashboard as well as to help customers see how they gain value from OIT’s services, he said.


Alvy Lebeck asked for a sample of metrics beyond the “standard” set.  Lew said that the metrics team worked with the messaging team to identify what information to collect through trial and error.  Specifically, the team wanted to identify what decisions need to be made and what information would be needed to make those decisions, he said.  Alvin asked if the metrics are primarily quantitative.  Lew said quantitative measure have been the focus so far, but the group is working to identify relationships between calls to the Service Desk and response rates at the system level to help provide a qualitative measure. 


Klara added that there is a difference between collecting metrics and monitoring.  She said that Kevin Davis and Mark McCahill have done a lot of work on identifying and scripting qualitative measures – such as response time to retrieve an email message.  Klara added that OIT is working to add a functional analysis rather than just a technical output.  Lew said the metrics team aims to consolidate the various views into a single picture.


Susan Lynge said the metrics team’s focus has evolved.  The initial effort was focused on technical system performance.  However, the team is focused today on measuring and reporting on end-to-end services.  Susan added that the team is starting with the customer experience and then asking what metrics would be necessary to determine if services are successfully delivered.  The team wants to get the customer perspective involved through surveys and/or focus groups. 


Paul asked if other organizations are already doing this.  Lew responded that most institutions don’t look at end-to-end service measurements.  John Board added that some graduate students have done work showing that measuring the user experience is still considered “novel”.  He added that all the systems may report full uptime, but the user is still experiencing problems. 


Mark McCahill said that OIT’s effort is generating some new ways of looking at services to measure their availability.  Alvy noted that service monitoring tends to have a bias since systems are monitoring for problems.  Alvy asked if there was any way to capture positive service experience as well.  Lew agreed that focusing on the negative measures is very easy.  The Metrics initiative is designed to watch capture the whole picture, he said. 


Susan added that the end-to-end approach of this metrics gathering initiative has resulted in a periodic cross-functional review of the data.  These meetings have been very helpful for participants.  Terry suggested the team look at web services and web presence from a business perspective as alluded to by Dr. Lange’s previous comments.  He added that web presence is crucial in recruiting graduate students.  Dave R. asked if there is a way to measure a mean response time for any given service for individual requests.  Terry said some web sites may not be visited at all so measuring response time may not be helpful.


Alvy asked the team’s status on measuring value.  Lew acknowledged the difficulty of measuring value, especially since it is frequently at the individual level.  For example, having very fast response time may be valuable to the individual, but it is very expensive to Duke.  Lew added that value perceptions differ per service.  Tracy asked when the group might be ready to report concrete examples, and Lew suggested back within a few months.



Changes to Port 25/SMTP Configuration - Klara Jelinkova


Klara said that the mail gateways sit at the edge of the Duke network.  These gateways handle traffic for 133 domains, including the Health System.  This edge filters traffic for viruses, both incoming and outgoing.  Klara said outgoing scans are important to reduce the possibility of Duke being blacklisted as a spammer by other organizations. 


ITAC discussed the merits of a proposal to block port 25 on certain network segments, and unless it originates from a known good gateway. Groups could add their mail servers to the outbound gateway through the Service Desk. 


Terry added that port 25 has been blocked on the 152.16.x.x subnet for five years. 


Bryan Fleming asked what might potentially break or what the impact would be with this implementation. Terry stated that when port 25 was blocked on the 16 subnet, breakdowns in communication resulted in some groups having issues sending mail.  Klara stated that this effort would be accompanied by a broad communication campaign rolled out after the end of classes. 


ITAC endorsed the proposal.



New Webmail Directions - Mark McCahill


Mark McCahill mentioned that OIT is looking to upgrading Webmail and Webfiles.  An impetus for this is the advancement of web browsers in support of AJAX and other technologies to provide a richer web experience.  In addition, these systems can be skinned to optimize with a given web browser or a given work style.  The current old Webmail implementation is based on IMP, he said.  After evaluating several different alternatives, the technical group selected atmail (http://atmail.com/webmail-client/). 


The student focus groups that participated in the evaluation liked atmail.  Atmail also supports running in other modes for backward compatibility.  The various modes enable drag-and-drop, integrated contacts list, and more. OIT is looking to move Atmail to be the default Webmail experience, while still allowing IMP for those who prefer it.  Tracy asked what the timeline would be.  Mark said Atmail would be enabled as the default at the end of the semester.


Mark said that Webfiles is a web interface for the AFS space.  It is much easier than installing an AFS client, he said.  In addition, you can access your files from any browser.  Webfiles was introduced in 2007 and due for a refresh since it does not support the AJAX stuff. 


Mark said that a Duke student working with OIT identified possible improvements to the Webfiles interface.  Mark demoed the current Webfiles along with some of the potential improvements in the new interface.  One Webfiles limitation is the single file manipulation.  Mark said using WebDAV circumvents the browser limitation.  Mark said that the intended implementation timeframe for Webfiles is the end of the semester.


Mark next demonstrated a beta integration of the jabber chat tool with web applications.  He showed a sample jabber chat window integrated with a Webmail window.  The development team experimented with enabling the chat window in any web application.  Klara added that the student participation has been a very creative environment that has helped improve the interface. 


Bryan suggested that OIT could possibly recruit students from software design courses.  Alvy suggested that this model could potentially replace the Blackboard dropbox.  Bryan asked if jabber integrated with other IM protocols.  Mark said it is its own, but federated, enabling other groups, such as Google.