Last week, I showed a prototype of a thin client solution to my district tech team. This particular test was just to show proof of concept and to get some feedback from the team on whether this is worth pursuing. I installed the K12 Linux Terminal Server Project on an old 800 mhz P3 with two network cards. That’s nowhere near enough horsepower for a production server, but I just wanted to give everyone a feel for how the system would work.

Before the demonstration, I did a little hand-waving to make it appear to be a little more robust than it actually was. For starters, I copied the security files from our regular servers over to it. Suddenly, everyone had an account. I set up some nfs mounts, so they could see their home directories, and then they could log in. These are things that could be automated and improved, so I didn’t feel like I was misleading people too much.
I wanted to do some customization to the user interface. For example, I wanted to specify the proxy settings and default home page in Firefox. I know, I can do transparent proxying, and I will, eventually, but this was an easier way to go. I also wanted some shortcuts on the desktop, some items removed from the menus, and a few other little configuration things. So I made a customized user profile, and edited /etc/profile to copy these settings to the user’s home directory upon first login. This makes the first login take a little longer, but it ensures that everything is pre-configured and ready to go. I also added a post-it note to the default configuration welcoming the user and explaining what this is.
The most obvious choice for a pilot project for us is a middle school computer lab. Currently, we have three locations at the middle school where there are classroom sets of computers. The 7th grade computer lab is used by the 7th grade computer applications classes, and by a couple 6th grade classes taking keyboarding. The other computer lab was intended to be signed out by teachers for use with their classes, but several keyboarding classes in the lab make this nearly impossible. The third location is the media center, which is always busy. If we could take some old machines that are being replaced next year and use them as thin clients, we may be able to build a limited-purpose lab just for the keyboarding classes. This would take the pressure off the other two labs, and make it much easier for teachers to sign out labs to use with their students. Because the scope of the lab is also pretty limited, it would be a good way to start using LTSP without everything having to work all at once.

In order for this to fly, we would need the keyboarding software to work. I was surprised to find that the package (Ultrakey) works fairly well with the Wine Windows emulator. It also looks like Accelerated Reader will work. We do have some graphic artifact issues, but nothing that’s a dealbreaker. It was a little disheartening to find that the entire Wine installation, including all of the Windows applications, are stored in the user profile. After moving this to another location and creating symbolic links, it made the logins much faster and saved a lot of disk space.

It would be really helpful to have an office productivity suite, and the team was impressed by OpenOffice. We browsed public drives on the various building servers, and flawlessly opened random Word and Powerpoint files.

In the end, the things that I expected to be concerns weren’t. They didn’t care that the computers weren’t running Windows. They didn’t insist that we try to run MS Office in an emulator. They weren’t overly concerned about the lack of USB support, or local drives (CD, floppy, etc). I know that some of these problems can be overcome, but I told the team that local device support would be very limited. I’m more worried about USB flash drives not working than they are.
They were concerned about sound. I’m going to have to work on that. We also found that the machines we ended up using as clients (Dell Optiplex GX260’s, because they were handy) didn’t autodetect the graphics cards, so we were in 640×480. That won’t work, but it’s hopefully just a driver issue.

The next step is to put a prototype in place in the middle school. Let teachers and students play with it, and make a determination in early spring about whether we want to try this for next year. Then, we have to find a room to put it in.