Barrio Malawi - Fulbright Journal

VSAT Community Networks

jon — Fri, 12 May 2006 17:15:47 -0400

I just read that the internet is growing faster in Africa than anywhere else in the world. Most of us here in Africa are connecting via satellite or VSAT connections which are a fairly expensive way to transmit data (though costs are rapidly dropping). While I am finding sufficient knowledge on how to purchase VSATs, I have found very little discussion or collaboration online concerning how best to maintain VSAT internet connections.

Just yesterday I received an email from a friend here in Malawi describing the experience of another friend who “successfully” installed a VSAT connection for their NGO. The NGO found that the the installers from the company they hired to install the VSAT “haven't got a clue [how to install a VSAT]. It took them over 10 days to get it running - basically they installed it and it didn't work and they couldn't work out why.” After asking help from a local internet cafe owner the NGO found “the VSAT company had supplied the wrong kind of network cable (not a cross over one) - a fairly basic mistake and they haven't apologised or anything - so we're not really very impressed.”

Poor quality technical support seems to be a common complaint of those organizations installing VSATs. Often the price for poor technical support in VSAT service contracts is very high. Organizations are usually asked to pay travel, room and board expenses for foreign technical support consultants in addition to paying fees for fixing connection problems. In the case of the University here in Mzuzu, we have found that maintaining the connection can be challenging because the network infrastructure was designed by outside consultants. Very few of the skills needed to keep the VSAT running optimally were transferred in documentation or training to the University IT staff.

Here in Mzuzu there is some consensus that organizations using VSATs could themselves improve technical support services while reducing service contract costs by collaborating online. We are interested in creating an online community to help other organizations around the world sustain and manage their VSAT internet connections. This would include: tools/guides to help organizations estimate and budget for usage costs, a tutorial collection to help build staff capacity for maintaining the VSAT and reduce dependence on outside consultants, a collection of success stories and best practices/uses for internet (giving professors blogs, etc), an online forum for technical questions to be answered by other members of VSAT community.

The other phase of our work here involves designing open source software to control, allocate and optimize network bandwidth on the VSAT (as I discussed before). I truly believe that this is the this is the other piece of the puzzle that could really fuel of fire of internet development here in Africa. User friendly software that turns old PCs into network appliances for allocating bandwidth would empower community groups, NGOs, churches, universities, and cities and allow them to collectively purchase VSAT equipment and distrubute and manage their bandwidth according each organizations contribution towards the connection cost.

I have found that the NoCat (http://www.nocat.net) project is pretty much what we need to begin allocating bandwidth based upon user logins. It is a great community networking product, but the problem is, the project stopped development sometime in 2004. Currently assessing the state of the NoCat code to see if it suits our needs and can be built upon. We are also looking into rolling out our own system.

Here is a more technical details of the community network system we are trying to create:

Place a Linux authentication gateway to filter all network traffic just before it is sent out the VSAT to the rest of the world
The authentication gateway should optionally allow unlimited local network traffic (email, file and print sharing, web browsing)
When a user tries to send data out to the world through the VSAT, the authentication gateway captures the request and the user is asked to login
After logging in network bandwidth limits are set on the gateway on a per user basis according to which network privileges the user has been granted. These are settings controlled by the network administrator.
The user is shown how much network time they have remaining in their account, how much bandwidth has been allocated to them etc.
The user is redirected to the website the originally requested before the capture. A small javascipt frame in the web browser continually checks in with the authentication gateway so that the gateway knows the user is still using the network.
Once logged in the user can send any type of network requests that are permitted by the network administrator through the VSAT
To end the VSAT network session the user closes their web browser
The authentication gateway notices that the user has not checked in for some time and closes network access to the VSAT for the user.

NoCat does most of this already, but is lacking an accounting system that would permit easy creation of network access limits for groups of users. Now to figure out if we can add it to the system... I love nerding.

Professor Saints and the nutty open source textbook

jon — Sun, 23 Apr 2006 20:44:26 -0400

I must apologize for the lack of content on the blog recently. Things have been busy in good ways... very good ways. I have been asked to teach the first ever introductory to computer science course here at Mzuzu University Department of Mathematics! Fortunately the course is only meeting twice a week, so there is plenty of time to piece together lessons and lectures. So far, the students seem to be enjoying themselves. They are very excited about having the chance to learn to program computers. Our only problem so far is that the number of computers on campus comes no where close to matching the students enthusiasm for the subject.

Agreeing to teach at the University has also opened some doors for my research. The other professors seem much more comfortable talking to another lecturer. When I was considered to be only a researcher, it seemed like the faculty here wasn't sure where how to classify me or how they could really work with me.

For our class text book we are using a very neat book called “How to Think Like a Computer Scientist” (http://www.thinkpython.com). I read it this past summer and quickly found it to be one of the best introductions to computer science that I have ever seen. The examples are clear, and topics are presented in a way that gets students writing fairly complex programs in a short amount of time. In addition to being a very high quality text book, the book is also published under an open source license which means the authors have given rights to everyone to download, print, edit, rewrite, or reproduce the text book legally in any way they choose.

The book is has three main authors – a professor from University of Chicago, a high school teacher, and a professional programmer – and claims to have received contributions from over one hundred readers who have proof-read and improved confusing sections of the book. I know that some might ask why on earth the three authors have agreed to let their work be redistributed for free without requiring any sort of monetary compensation from the users of the textbook. You are welcome to ask the authors themselves, but I think the answer might be that by opening their book – making it available for free - the authors have, in fact, benefited greatly economically. Hundreds of people have proof read and contributed to their textbook and the authors have not had to pay any of them a cent! Could you imagine paying an army of one hundred people to edit and improve a book you had written? Also, by opening their book, the authors have built names for themselves. Without waiting for a publisher's approval, these authors have published their book and watched it quickly gain in popularity in a genre where there are hundreds of competing textbooks and authors.

I took me a while to explain the concept of an open source textbook to the faculty here at the University. In Malawi, they are used to receiving “free” textbooks. Loads of old, hand-me-down text books arrive from universities in the USA and Europe each year for “free”. These textbooks are “free” because they are old and no longer wanted. My challenge here was to explain that this open source text book is “free” for a different reason. “How to Think Like a Computer Scientist” is free because the authors have decided that the most productive way they could produce a high quality, cutting edge textbook is to give it away for free and encourage students and teachers around the world to improve the text. The textbook is “free”, not because its not longer good enough for the universities of the USA and Europe... but “free”, because the authors have successfully adopted the proven open source development model (which has brought numerous high quality software products to the market) for literature.

My students, like all university students around the world, are pretty short on cash. The high quality and low cost of this text book made it the ideal textbook for our introductory computer science course here in Malawi. We have printed 5 copies ($4.00 each in our library) for a class set for the library, and the students are currently saving electronic copies of the book to their floppy disks that they carry with them. The administration here seems to also really like the idea of the open textbook. Next year, when Mzuzu University wants to get its hands on one of the latest and greatest computer science text books the University won't be dependent upon the charity of foreign universities or publishers. The updated version of “How to Think like a Computer Scientist” will be there “free” and ready for our students to download and begin reading.

As I think more about the textbook, the way in which it was produced, the impact that it is having here already... I realized more and more... open source, as model for production, clearly stands to have a enormous impact developing world.

Yet I still believe that most of us, having grown up in a copyright society, have a completely wrong impression of what open source (copyleft) development really is all about. Open source production is not some trendy new fashioned form of communism... there are real economic advantages and efficiencies at play here. The way “How to Think Like a Computer Scientist” was able to “hire” hundreds of editors for almost no money is just one example.

My current (though humble) career is another. Most of you know that I spend quite a bit of my free time and earn the majority of my income by working on open source software projects. Yes, its true, I earn money by giving my software away for “free”. But, I want people to know that I give it away, not because I am a good samaritan... I give it away because it is the best way for me to make money. I am an efficient programmer. I am a not such an efficient marketer or a lawyer. If I tried to market my software or enforce copyrights upon the code I have written, I believe that I would be wasting my time, money, and energy. I would rather spend my resources writing more creative software. By distributing my quality code for free I encourage people to use it. The more people that use it, the more likely one of two things will happen: one of the users later be willing to pay me to improve the code I have written, or some other programmer will add functionality to the code then share the changes back with me. Whether the a new client pays me to write new code, or a programmer (that I don't pay) writes a new feature that I can then sell to my clients... either way, I gain economically.

We should be clear that producers (like me) aren't the only ones that benefit in the open source production model. Buyers (my clients) can benefit as well. Usually when I am considered for hire on a software project, I tell the prospective clients that I will only take the job if I am allowed to give away the code that I write during the project for free to the open source community. Clients are always nervous about the idea at first. But there are strong economic arguments they must consider as to why it is to their benefit to pay me to give away my code for free. Sounds crazy I know... but take a client that recently hired me for a small $1000 USD dollar development project in Washington DC as an example. For that price I was able to develop some new functionality for their website. The problem was that I would soon be moving away to Malawi and no longer be able to help them maintain the new functionality. After I completed my work, we “open sourced” the code I wrote. We contributed the code to one of the major open source software projects called Drupal (software to help people manage websites). Shortly afterward, I began receiving request from around the world from other people who were interested in using it in their client's websites. Whether these people sold the code I wrote to their clients or not, it didn't matter. We let everyone who asked have the code. To every one I sent the code, I included a note said I could not promise to maintain the code after I moved away. They would have to use/sell/study the code at their own risk (this is common in open source production). Just two months before I left for Malawi, one of the “free lunch” users contacted me with the news that they were interested in taking over development of the project. They would to continue to update the code and add even more new features to it (not so “free lunch” after all).

Just last month I noticed that a new version of the code I wrote was released for everyone to use for free. There is now a team of three programmers working to maintain the code. The code is higher quality and includes even more new features than the original code i wrote. My clients in Washington DC are now free to use this new and improved code. So for the small price of their $1000 original investment my clients also now have a team of three programmers working to help maintain this new functionality for them that they do not have to pay. They get all of this because they were willing to take the risk open sourcing production of product that, in the end, proved to be valuable to other people.

And so, now I have to admit, that I really do not like the first chapter of the book “How to Think Like a computer Scientist”. I am thinking of rewriting the first chapter of the book using material from the first lecture I gave for class here entitled “How to think like a Computer”. If I do write the text, I will give it away for free to the authors to use in the next edition of the textbook. Why do I want to write a chapter for the book for “free”? Yes, there is part of me that just wants to see a better version of the book next year... but also I know that it would benefit me and my resume greatly to be able to say that I was a contributing author to one of the best computer science text books out there. If I do write a better first chapter both the book's readers, and myself stand to benefit. This is how open source production works.

Differences between internet connectivity in Colorado and Malawi

jon — Sun, 23 Apr 2006 20:37:34 -0400

Malawians, in general, are both very poor and very ambitious. I think this is why I am finding engineering in Malawi to be such exhilarating experience. So often here we are asked to do a lot with a little. Its forcing me to push the limits my creativity, my imagination, and my skills as an engineer.

I was in a meeting last week to discuss the redesign of the internet network architecture for Mzuzu University. There is a need to rethink how users are allowed to connect to the internet here in Mzuzu. Currently the network is designed much like the networks of the universities in the USA. In general, computers with connections allow users are to have unlimited access to the internet. A quick comparison of the internet connection here in Mzuzu to the last connection I used at my house in the USA will show why a different network architecture is necessary in Mzuzu:

	Mzuzu University	My Mountain House in Colorado
Bandwidth (speed of connection)	256 kbps	> 1000 kbps
Number of users the connection serves	about 500	2
Price per month	at least $300 - $400 (this could be much more, but getting the actual numbers has been tough)	$55

In English the table basically says: In Mzuzu we are trying to serve about 500 people with an internet connection that is about 5 times slower than the one broadband users have at home in the USA. This same slower connection costs us about 6-8 times what it costs home users in the states.

Again, the trick is to do a lot with a little. At home in the USA where there is a lot of cheap extra bandwidth readily available the efficiency of the design of your home network is not of much concern. If for example there is a virus sitting on your home computer sending out data via your internet connection, chances are the data the virus sends costs you very little money. Here in Malawi this is not the case. Over our satellite internet connection, bandwidth is scarce. Viruses sending out data cost us a lot of money. For many reasons of both, economics and engineering, we need to be very efficient here with how data is sent from our network out to the world via the internet.

So Mzuzu University has put together a team of myself, a volunteer from Japan and the university IT staff to design a system network architecture to use the resources of the VSAT internet connection more efficiently. We decided unanimously that we would try our best to carefully document the system we design so that it could be easily implemented by other universities who use satellites to connect to the internet.

Overall our goal is to design a system that gives as much free access to the internet as the University can afford. We want to limit users only so that they realize their internet time is expensive and should be used only in the most productive ways. By limiting the number of hours users have available for free they will respect the time available to them more and will begin to treat the internet as a limited resource.

In our first meeting we decided the requirements for the new university network. I am posting them here in hopes that some of you computer networking nerds out there might take internet in the project.

The requirements :

The network should allow users unlimited access to the local network (intra-univeristy network traffic like email , file sharing, and web browsing cost us very very little and should be allowed to happen all of the time)
Only network traffic sent via the VSAT to the rest of the world should be limited (this is the expensive network traffic)
the University has decided on set number of free hours for various groups (staff, students, administration, etc) available per week or per semester and on a pricing schedule for additional internet time. The network should track individual users use of the VSAT connection and limit their connectivity based upon the limits defined by type of user they are ( for example, a student may be given 3 hours of free connectivity per semester, while a staff member may receive 5 hours per week).
The pricing architecture should allow administration to change pricing of internet access easily and frequently
Users should be able to view their reaming time and account balances
Users should be notified with their remaining free time or account balance is low
System should allow large files and full mirrors of websites to be downloaded during off peak hours for internet usage (nights and weekends) at no charge to University network users
When power goes out the system should automatically credit one minute to users accounts and stop deducting time
new network should continue to use the web proxy cache already in place to minimize duplicate downloads
new network should include bandwidth optimization and local queuing of internet traffic to allow highly interactive traffic like Voice over internet, Video conferencing, ssh, web browsing to take priority over non-interactive uses of the internet connection like file downloads (this will also discourage large number of large files from being downloaded as they will go at slower and slower rates without affecting interactive web traffic)

Thoughts anyone?

One thousand roads and no highway...

jon — Wed, 22 Feb 2006 05:57:19 -0500

Three days into our Malawi adventure and my head is full of questions. I think that I have pieced together a reasonable view of what the internet infrastructure looks like in Malawi through various meetings at the US embassy here in Lilongwe.

It seems that in Malawi, any government office, international organization or private company who wants to connect to the internet does so by buying their own VSAT satellites. Its essentially every person for themselves, as each organization takes their chunk of capital and builds their own personal pipeline.

Its as if a village of people wanted to travel to a distant city, instead of pooling resources to build a highway that would provide faster/more reliable/efficient transportation, the people each decided to carve their own individual narrow roads to the far away city. There is an unbelieved able amount of upkeep and investment costs associated with keeping these "individual roads" or fragmented internet oasises running.

This fragmentation is frustrating to me, again, I am new here, but it seems that these VSATs are very sort-term and individualized solutions that serve very few people. They could never really be the economic catalysts to Malawi that I think the internet should be. Maybe I am naive, but It makes me wonder if some sort of resource pooling campaign couldn't be possible to link up to the nearest fiber connection here in Africa?

There is a rumor here in Malawi that there is no wire that crosses the Malawian border... not power... not electricity... not data.

I wonder where the nearest fiber optic cable is to Malawi? Who could I begin asking? Could it be streched here?

Fulbright Research Proposal

jon — Tue, 21 Feb 2006 01:33:42 -0500

Here is a shortened copy of the proposal I wrote for the Fulbright grant. It describes the project I will be working on while in Malawi

Overcoming the lack of access to appropriately implemented technologies in the developing world is fundamental to the problem of realizing social development. By increasing productivity and diversifying the workforce, the proper use of technology lays a vital foundation for public health, democracy and economic development. This is certainly true for the proliferation of information and communication technology (ICT). Studies by the Digital Opportunity Task Force, the United Nations Development Program, the World Bank's Information Development Fund and Harvard eReadiness Project all prove that the expansion of ICT connectivity can augment social development. Building community telecenters, enhancing rural commerce via ICT-based microcredit lending, launching web-based e-commerce, using the internet to teach public health and land management in rural areas, and using email to relay commodity prices: these are among the many ICT strategies that can enable faster progress. Increasing access to technology and crossing the digital divide is not an option; it is an absolute necessity if the poorest nations are not to fall even farther behind.

When I came to the University of Arizona in 1999, I was intrigued by a project now known as the Broadband for Development Initiative (BDI). BDI is an effort of the University of Arizona and Mzuzu University (in Malawi) to augment economic development in Northern Malawi via the use of ICTs. As I worked on the project, my curiosity grew into an intense desire to work in Malawi. This past summer I was fortunate to experience the country and its diverse culture, a place where I have been dreaming about working for years.

I arrived in Malawi at an exciting time for the BDI project. After three years of negotiations, Mzuzu University and three other national universities had acquired VSAT satellite systems that would finally provide affordable and reliable broadband Internet access to their campuses. For two weeks I traveled throughout Malawi and participated in discussions about the new system with intellectual leaders, Mzuzu University administrators and faculty. Nearly everyone in Mzuzu expressed excitement over the coming technology. Those that I spoke with also expressed an urgent need for research to determine culturally compatible methods of integrating broadband access into northern Malawian society in order to provide the greatest potential for economic development.

This project aims to create a comprehensive report to guide Internet for development efforts in the Mzuzu region of Malawi. My research will adopt a unique and proven method of creating sustainable Internet based development projects used by Dr. Barron Orr at the University of Arizona known as “stakeholder-driven development”. The outcome of the process will be a report that provides Mzuzu University administrators with recommendations of specific ICT for development projects that would be sustainable and most beneficial in the Mzuzu community.

In stakeholder-driven development, those that are meant to benefit from Internet applications are involved in their creation from the beginning. The first step is to assess stakeholder needs through participatory rapid appraisal techniques that will provide a full ethnographic assessment of both the individual user and community needs, information and connectivity gaps. This phase of the project, which will combine focus group and key respondent interviews and a quantitative survey instrument, will last for four months. I will interview business owners, local leaders, University staff and townspeople in order to become familiar with the needs and goals of the Mzuzu community. By the end of four months I hope to have an accurate picture of the community's most pressing needs and aspirations, including a short list of “lead users” or early adopters who will participate in the applications development phase. Based on the results of the needs assessment, I will narrow the scope of the project by selecting one or two specific development needs and target populations that proper application of Internet technology could help address.

The next step of the “stakeholder driven” development is to create prototype applications in collaboration with colleagues and students at Mzuzu University that can be tested and critiqued by the lead users identified in the needs assessment. I plan to devote months five and six to creating the prototype systems. Prototype systems will be created by leveraging the knowledge and experience of the “best practice” projects and adapting them to meet the needs of the Malawian target population. The prototype systems will be simple staged demos of what a full-featured system would entail. Each prototype will address a specific community development need as determined in the needs assessment phase of the project. Some possible projects include: enhancing field work activities (natural resource management, public health, agricultural extension) by making web-based information products mobile through hand held Personal Digital Assistants, helping University professors integrate the Internet into their classes, or working with the local hospital to improve record keeping infrastructure.

In the third and final phase of the research, lead users (and later the broader base of stakeholders) will be invited to a series of guided workshops for testing of the prototypes. Their opinions and experiences, difficulties and suggestions will be meticulously recorded. Their feedback will be used to identify and prioritize ICT for development projects best suited for the region. Based on the ethnographic research of Phase 1 and the systems research of Phase II, I will compile the stakeholder recommendations along with the corresponding prototype systems specifications into a actionable report for University administrators by the end of month ten.