Draft Trip Report, 19 January 2008

This is the draft of a trip report that I wrote in January 2008, following a visit to first a conference at a HCBU (Historically Black University or College) near Dayton, Ohio, that was then followed by a visit to a small company in Detroit. The company was then known as ACES, and is now QWK2LRN.

I sent a copy of the finished report to several IBM executives, but without any success.


David Shields/Watson/IBM

01/19/2008 01:35 AM

To

jim hare

Subject


draft trip report

First draft of my trip report. Comments and suggestions welcomed.

I recently attended a conference for black educators in Xenia, Ohio, and then spent two days in Detroit to learn about a small company called ACE. This all came about from my desire to learn more about a company named ACE, to follow up in my meeting their CTO, Jim Hare, at a conference in Indianapolis this past October for educators on the use of open technologies such as open-souce to improve education.

— Add part about Rudy and Ramon here, as well as name some of the people and groups I spoke with —

[ Added 18 Mar 2009: Rudy was an IBMer and volunteer interested in education; Ramon works for a nationally known group for executives that includes at least one IBMer on its board.]
The issue at hand is how to improve the use of computers to support education in large, poor urban school districts such as Detroit.

Here is what I learned during my trip.

Computers currently play no meaningful role in providing education to the vast majority of the students in the Detroit Public School system. The average computer is a desktop close to ten years old running Windows 98, and one of every two or so is broken. Those that do work are not reliable, and there are not resources at hand to diagnose and repair those that fail in a timely fashion. Schools have closets filled with computers that no longer work, or that sit in boxes that have never been opened.

Since there aren’t enough computers, and those that do work are not reliable, teachers are not willing to commit to using computers to play a substantive role in providing education in the classroom.

Detroit schools do have access to high-speed internet, and the city pays a substantial amount to maintain this infrastructure, yet it goes largely unused.

This situation is not confined solely to the Detroit school system, but occurs in many of the nearby school districts, some of whom accept students from Detroit. The state of Michigan funds students attending districts other than that in which they live, so there is an incentive for school districts with underused capacity to find ways to attract new students.

Detroit educators are well aware they are on the wrong side of the “digital divide,” and the situation in Detroit, as well as Michigan in general, is so dire that there is a willingness to try new solutions that offer promise to improve education, particularly to meet the requirements of the No Child Left Behind (NCLB) legislation. For example, one school has changed math textbooks three or four times in the last few years, each time hoping that the newer textbook would somehow improve math test scores. The cost of switching textbooks is a major expense, as some textbooks now cost close to a hundred dollars a copy.

The drive for change comes from the educators and parents. The people providing the current IT infrastructure have a vested interest in maintaining the status quo, and are the ones least interested in trying new technology. Attempts to market by selling to this group by promoting a technology on its newness are thus doomed to failure.

The most important lesson I learned during my trip is that to today’s younger educators, and to the education students who will be our future educators,

THE COMPUTER IS A DEVICE FOR DELIVERING THE INTERNET TO A CLASSROOM DESKTOP

THESE USERS DO NOT CARE IF THAT DEVICE DOES OR DOES NOT RUN WINDOWS. They just want access to the internet via a browser on a device they can rely on.

Today’s teachers do not want computer labs. They want one-to-one computing, with the internet on every student’s desktop in their classroom, not in another room they have to go to.

The question then is how to reliably deliver the internet to a classroom desktop at the lowest possible cost?

Here is how ACE proposes to do it, based on their work over the last several years.

First, using the current approach of providing a standard commodity PC is too expensive. Moreover, maintaining the integrity of the device is an important issue. Laptops are out because they are too fragile. New computers provide more function than is needed, and will likely be stolen. I heard numerous mentions of school breakins where working computers were stolen. Theft is not always done by outsiders; theft by staff is common. In one extreme case, the computers in a school were brought down because thieves stole the copper wiring from part of the power lines going to the school!

LCD displays are also suspect. They are prone to theft because of their light weight, and they are prone to damage by students who, for example, point at the screen using a pencil or pen which comes in contact with the display surface, which is easily scratched. Thus, urban schools such as Detroit favor the usr of CRTs. Even though they take up much space on the student’s desktop, they are sturdier than LCD displays, and their weight, as well as their old technology, makes them much less prone to theft. Also, most schools have a ready supply of old CRT’s, many of which are in good condition because they have been so little-used.

ACE takes old commodity PC desktops, either from the school or from a supplier. They refurbish them by removing the hard drive, cleaning them up, and making sure they can be booted up over the network. ACE does not support floppy drives or CDROM drives since they are not needed.

These refurbished machines consume one-third less power because the hard drive has been removed. The device runs much cooler, and is much more reliable than a standard desktop in that the possiblity of disk drive failure is not an issue.

ACE offers these devices at a one-time cost of $100, and then charges $100 per device per year, committing to replace devices that fail and to provide internet access to the devices.

Note also that recent advances have made it possible to provide similar function by small boxes that can be held in one’s hand, and can thus be bolted to the underside of a student’s desktop, at the same cost. These devices are not a target for thieves, as they cannot run standalone. They need a server to power them.

The question then is how to provide the necessary server?

Now all that I have just said is nothing new. The refurbished diskless PC is what is called a thin-client, and there exists software, in the form of the Linux Terminal Server Project (LTSP), that allows a Linux server to control these devices. For example, LTSP is the core piece of software in the Ubuntu variant Edubuntu.

The problem is that LTSP is hard to install and configure. More importantly, LTSP DOES NOT SCALE.

Indeed, this is the main problem that ACE has been working on for the last few years, that of scaling up thin-clients to work in the largest, poorest urban school districts.

ACE runs a modified version of Red Hat Enterprise Linux (RHEL) on the server. Their solution can support many more client devices than can LTSP. Also, though most users are not aware of it, the student’s desktop is a full-function Linux computer. Students and teachers can run Perl, Python, Gimp, Open Office, and so forth.

The student’s device is quite powerful. A refurbished ten-year old commodity PC running Linux can deliver better performance than can a current machine running Vista. For example, while attending a demonstration of the ACE technology
at the conference in Xenia, I suggested the following experiment to the twenty plus participants: Bring up Google in the browser on your desktop and enter a search string, but don’t hit Enter until told to do so. I then asked Jim to count down to zero, at which point everyone hit Enter. I was sitting in the back row, and so could observe that each screen was updated within two seconds. Also, the student sitting next to me spent much of his trial time viewing videos, and mentioned that the performance was better than in his own school, where he often had to pause so the computer could catch up.

A good-sized xSeries server can power quite a large number of student devices. However, this solution, while quite
adequate, poses a problem: how to maintain and support those servers? While it is an easy task to replace a failed client device with another refurbished PC, server support requires real technical skill. You don’t want to have a dedicated support person in each school of district.

The question then is can you support the student devices using servers that are NOT located in the school, or even in the school district?

ACE takes as their reference point the following school system, that of New York City: one million students; 1200 buildings; 60,000 teachers; and hence 60,000 or so classrooms. They have worked to build a solution that can scale to this district, not just much smaller ones.

Does there exist a server so powerful that it could support a school system this large? Yes, there is. It is called an IBM mainframe, and IBM has spent over four decades learning how to build and deploy it.

Indeed, this is why I took some vacation time in January and paid for this trip out of my own pocket. I wanted to see if this possibility made sense.

I now believe it does, and so presents an exceptional opportunity to IBM.

The opportunity is to use IBM mainframes to deliver this workload to tens of thousands of these client devices on student desktops.

This workload is especially attractive. The server is, for the most part, just receiving HTTP requests from the students, and then delivering the response in the form of HTML; the browser in the client device does the graphical rendering. Also, the workload is linear. There are no critical points such as simultaneous access to a database. Adding support for an additional two thousand students should cost just twice that for adding one thousand. If all the students in one class are viewing the same chapter in a 19th century novel via the Gutenberg project, then caching is our friend.

This is a case study in server consolidation: move the servers in the district to a large IBM data center. It is also a study in staff consolidation. All the sysadmins in school districts who, for the most part, are teachers drafted into this role because they have an expertise in computer technology, can be replaced by a handful of very skilled experts in the remote data center, and thus can go back to teaching. This is especially important in that teachers who are skilled
in computing, as are teachers skilled in mathematics and science, have many job opportunities outside schools: we cannot rely on dedication alone to provide high quality sysadmin-type skills within a school.

Current estimates are that it costs about a thousand dollars a year to maintain a commodity PC desktop within a school. The ACE client device costs one hundred dollars a year to maintain. That leaves nine hundred dollars. How much of that would be needed for a solution based on the mainframe? I would venture less, much less.

There is another advantage to the mainframe approach, though it also applies to the individual server within a district. While the primary use will be delivery of the internet, some teachers, and also adminstrators within a district, will have legacy applications they would like to run under this new model. This is no problem, as we can just provide virtualized instances of Windows/95, Windows/98, Windows/2000 and Windows XP on the server that can be accessed from the classroom desktop. For example, the server could support Reader Rabbit running under Windows/98.

The question then is how to pay for all this? There is an answer, in the form of two of my favorite phrases of late, “Title I” and “e-Rate.” In brief, these are federal programs that provide funding to qualified school districts, with the most funds going to the poorest districts. These funds have already been used to bring the internet to most school districts in this country. The problem is that the current approach to deliver internet access using full-function commodity PC’s is both too expensive and has other problems, as noted earlier.

Simply put, the cost to the poorest districts of moving to this model consists of just the cost to refurbish and support old PC’s that are not currently being used, and the cost of running the mainframe servers. The server hardware itself would be funded by Title I and e-Rate and would require no funding from the school district.

This is not just about k12, it is also about universities, especially the Historically Black College and Universities (HBCU’s) such as Ohio’s Central State University, the host of the conference in Xenia.

This is also not just about schools. For example, a large automotive company in Detroit, as well as the State of Michigan, currently have tens of thousands of PC’s that could be supported using this model.

If IBM can’t figure out a way to partner with ACE and use this model to deliver the internet to students and teachers in districts like Detroit, as well as other similar institutions with excess old PC’s that are not being used or are costing them too much to maintain, then all I can say is, “Shame on us.”

thanks,
dave

Dr. David Shields

Program Director, Open Source Software

Linux Technology Center, Systems and Technology Group

Advertisements

2 Trackbacks

  1. […] The Wayward Word Press WordPress gave me a blog to see how I was going to handle it « Draft Trip Report, 19 January 2008 […]

  2. […] You Have a Problem March 19, 2009David Shields, daveshields, is looking for a job March 18, 2009Draft Trip Report, 19 January 2008 March 18, 2009Dave Shields, daveshields, Resume (18 March 2009) March 18, 2009Why Being Open is […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

  • Pages

  • March 2009
    M T W T F S S
    « Feb   Apr »
     1
    2345678
    9101112131415
    16171819202122
    23242526272829
    3031  
  • RSS The Wayward Word Press

  • Recent Comments

    russurquhart1 on SPITBOL for OSX is now av…
    dave porter on On being the maintainer, sole…
    daveshields on On being the maintainer, sole…
    Paul Tallett on On being the maintainer, sole…
    mrrdev on On being the maintainer, sole…
  • Archives

  • Blog Stats

  • Top Posts

  • Top Rated

  • Recent Posts

  • Archives

  • Top Rated

  • %d bloggers like this: