Duke ITAC - April 25, 2013 Minutes

ITAC Meeting Minutes

April 25, 2013, 4:00-5:30

Allen Board Room

  • Announcements
  • AFS/CIFS (Mark McCahill, Richard Biever)
  • Digital Signage (Steve Toback, Jack D'Ardenne, Teresa Jennings)
  • Kuali OLE (Molly Tamarkin)
  • Reception


Ashutosh Kotwal announced that this would be the last meeting of the semester and that we would move from a bi-weekly schedule to a monthly schedule.


Mark McCahill reported on the AFS to CIFS migration and data retention. He started by reminding us that back in 2012 the Common Solutions group agreed that long term AFS viability is dubious and that universities should start migration planning.  Because of this, Duke started a step-wise migration out of AFS. This is not a small task because we have about 10 terabytes of AFS storage. About 7 of the 10 terabytes are home directories. About 3 of the terabytes are in shared spaces, typically course space and some project space.

Step one seemed rational, to move the home directories that would minimize the exposure the fastest. This is what we’ve been doing for about the last year. We put together a migration tool that we call Portage. From July through the end of November and into December we had self-service opt in pilot migrations. This allowed people to migrate their materials back and forth if needed. That went well enough that we decided in December that we would do a mass migration of all of the active netIDs. There was still an option for anyone who had problems that allowed them to go back to AFS if they needed to for their home directory.   We also started making all new people only have CIFS home directories and not have AFS home directories. This went reasonably well so in February we disabled AFS home directories for all inactive netIDs. This allowed us to identify if they were in use. We’re carefully turning things off and turning them right back on if necessary.

As of today there are still 73 users with home directories on AFS. We’ve moved just under 44 thousand home directories for people with active netIDs from AFS to CIFS. At the end of that move more than 48 thousand inactive netIDs AFS home directories were de-activated. We can turn them right back on again if we need to. To date we haven’t had any come back with a problem. We’ll give it more time before we conclude that clearly nobody is using that feature.

There was discussion noting that many of the accounts, active and inactive, have never been used.

We have moved heroic numbers of home directories. There are still some stragglers who have good reasons to stay on AFS. We are going to work with them to sort out a strategy for eventually getting them migrated out of AFS as well.

Richard Biever addressed data retention. We are moving CIFS into alignment for what we have for other services such as email, network access, Sakai and others. Basically when we turn off your netID we start a clock. Our retention policy is that we will save data for 30 days after your netID is disabled. When faculty leave Duke we remove access after 30 days. In effect this means that it is 60 days before the data disappears. For staff it’s 31 days. For students it’s 395 days. For sponsored accounts it’s about a month. We do retain the data for OIT-maintained services in a disabled state for these times before the data goes away.

We still need to do some cleanup. We still need to identify the remaining users for upcoming cleanup.  We will contact the remaining 73 users that still have directories and talk to them about migrating into the new space. Eventually we will delete the disabled AFS home directories. The target date is September 1, 2013.

Mark explained the reason is that it will have been six months and people will have been through a semester. We have a theory that people may be doing research things that they may not be doing during an active school year and may be doing during the summer. This lets us cover that possibility. We would like to free up the space. Robert Wolpert pointed out that this is only a couple hundred dollars in discs. Mark and Richard explained that the cost is in the discoverability and the fact that it’s distributed over a variety of disc spaces that drives up the cost. It also simplifies IT lives a little by eliminating a whole technology. This allows us to focus on new technologies.

The new thing that is being worked on is NFS. The problem with NFS to date is that you have to trust both the client and the server to be secured. Tracy pointed out that AFS, NFS and CIFS are all ways to access a centrally hosted file system. Mark noted that to date we haven’t used NFS because we couldn’t secure it properly. We think we can with NFS version 4. NFS would be offered with Kerberos authentication in addition to CIFS. This would allow two separate systems for central storage interfaces. These would not be synced, but we may be able to sync these two at some point. The first step is to get the two spaces working properly.

Discussion followed the presentation. It was first pointed out that there is work to be done to clean up documentation prior to fall classes starting. We will be shutting down personal home directory services in AFS. There will still be shared spaces for courses in AFS. This will be a separate project to deal with.

Ashutosh asked how long we expect this technology to last. Mark said that we think at least 10 years. It’s not going away any time soon.

John Board noted that CIFS is a great personal file storage system. AFS was a file system that would do personal storage and shared storage in very flexible ways. The CIFS migration broke processes for those depending shared storage workflows. It’s not a one-size-fits-all solution for technical departments with these needs. There was discussion about addressing these needs.  

Tracy asked how large is the issue is for these. Is this just the 73 users? The groups in Pratt were also noted. Mark said that the reason for examining NSF version 4 is to address these 73 users as well as the Pratt courses.

Digital Signage

Steve Toback started the Duke Digital Signage presentation. Digital signage is essentially bulletin boards that can plug into the wall. Basic digital signage is digital flyers. It can also be more, including video, audio, dynamic news and interactive way-finding. We’ve bought a tool from Four Winds Interactive that can do all these things. Steve showed a display with an example layout. This system can also be used as an adjunct to the main emergency communications system in place at Duke. All screens would be required to be subscribed to a system so that emergency notifications can appropriately take over the whole screen.

This system is run by a single server at OIT on a VM. We did an RFP last summer with involvement from around the school, led by Tim Bounds from Student Affairs. We selected Four Winds Interactive, which allows us to push content to the signs from a central server. Transmitted content plays even if contact is lost. The site license covers all Duke locations including Beaufort and Kunshan.

Departments already using Powerpoint should be able to easily transition. There is a specification for an HD resolution jpg. We’ve built a template that will automatically add the weather, time and a ticker. The cool thing is that you can do more. Steve showed how a set of applications that can be added and customized.

Steve showed how users can create their own setup using a wireframe where you drop applications into boxes. The simplest application allows a user to drop in a Powerpoint slide. It is also possible to do a “content feed”. The content feed allows users to update content via an excel spreadsheet. This method can also allow a user to modify things like duration or start and end date.  Steve showed an example where OIT birthdays are displayed via excel spreadsheet that holds information and automatically shows birthdates and service dates for OIT staff at appropriate times. These processes don’t require content providers to log into the system. They just update existing documents.

There’s also an RSS feed function. There’s a weather widget that comes with the application. The date and time could allow a world clock.

There are three things that are needed to participate. First you need people to run the system. You will also need the hardware and content.

For the people, you’ll need an integrator to install the sign and a technical administrator. We recommend a Mac mini for the hardware. There is a bundle available at the Duke Computer Store. Sign developer training is available, or users can make use of the templates. Content creator should be identified to create slides for local content.

The Mac mini needs to be configured for pure windows.  You need one player per sign. It is possible to split signal to feed more than one sign the exact same content with video over cat5. The Mac mini is affordably priced including support and maintenance. This can also be done with other computers. Steve’s group has published specifications. We have direct support with Four Winds. Four winds will work with you if there is a machine that meets specifications and has problems. We’re also looking at NEC interactive screen that has pc built in. We don’t have a recommendation on screens, but we do have minimum specifications. Don’t forget that you’ll need power, mounting hardware and an internet connection. We don’t advise wireless.

To sign up there is a form on the website. Steve introduced Jack D’ardanne who is our service manager and Teresa Jenkins just joined Steve’s team as the lead systems person on this project. Steve showed a list of participating departments. There are 25 departments that have signed on. They are in various stages of getting up and running. Currently there are 14 active screens across the schools. Jack is currently working with DHTS. They have been working on getting through firewalls and have a workable solution.

There was discussion about implementation, scaling and cost.

Kuali OLE

Molly did an informational presentation on Kuali OLE. Kuali OLE is a community effort. Molly showed a slide with the partners involved in the project. Considering that the group has not yet developed anything that can be implemented yet, it is impressive that the number of schools involved has been growing. New partners include Villanova, NC State and a consortium of British schools. There is a mix of smaller schools like Lehigh and bigger schools like Chicago and Duke. The idea is that we’re creating an open source research library system. There is currently an open source library system called Evergreen that is used primarily in public libraries. We’re looking at an IOS that supports research at the university scale. This means materials in all different formats and in every language that can be encoded. The architecture needs to be service oriented, flexible, and enterprise caliber. This system will need to be able to integrate with local identity management and accounting systems.

Molly showed a list of requirements defined for the system and explained some of the history of the project. Mellon started funding in 2008. It’s been a little tricky because of issues with non-profits competing with the for profit businesses in this arena. Mellon was ok with funding since they deemed that interest in this area was stagnant. They were interested in funding an investigation to see what could be done to change the marketplace in an area where interest is smaller and smaller resulting no progress.

Mellon is funding the coding and the partners are funding the core team. We plan to have our first implementable release in fall of 2013. Lehigh and Chicago are set to be early adopters. The group is focused on these implementations as success will be vital to the future of the project.

The Kuali Foundation provided a great governance structure. Molly showed organizational chart and talked through highlights. There is a board of senior leadership, a functional council that Molly chairs, a community development council and a technical council. The project funds a full-time project manager and a core staff of about 5 FTE.  We’ve outsourced coding to a company that is both in the US and offshore. This allows us to only pay for hours that are used.

The current system state uses Apache service Solar and Jackrabbit. We are really using the tools of the web, not tools of library systems. This is a really big change in that it matches electronic workflow.  We’re using state of the art IT tools rather than library tools for search and discovery.

We’re also partnering with a group GoKB. GoKB is an open source knowledge base for electronic resource information. This allows sharing electronic resources in a way that benefits the community. This could create the first ILS with an electronic knowledge base. Right now we are paying for a knowledge base.

Molly then explained the difference between a doing this through a community and the marketplace. The marketplace is essentially dependant on growth and market share. The community is about sustaining the development necessary to the group. For the community, the process is as important as the product because it has to evolve.

During question and answer Molly clarified that this system is all back-end, not end-user facing. It will be packaged with open source access. She also pointed out that right now we are paying both for development and for the current system, but that when implemented we will pay significantly less than the current system cost. Implementation at Duke is targeted for 2014 with a rollout in July 2015.