A Linux thin client for every child

The One Laptop per Child scheme sounds like a smart way to solve the digital divide, but NComputing boss Stephen Dukker thinks he has a better idea
Written by Andrew Donoghue, Contributor on

In September this year, the Macedonian government announced one of the most ambitious educational technology projects ever proposed: to provide 180,000 of its school-age children with computer access.

The former Yugoslav republic has latched onto the potential of IT to drag it ahead of its Balkan neighbours and committed to transforming a largely agricultural and industrial economy into a knowledge-based one.

According to Ivo Ivanovsky, Macedonia's minister for information society: "The Computer for Every Child initiative is the largest and most important education project undertaken in the 15-year history of the Republic of Macedonia... Our goal is to build a knowledge-based economy in which our entire workforce is educated using information and communication technology within the next five years."

Aside from the logistics of rolling out new infrastructure on this scale, the obvious question is, how can a country that hardly ranks as one of the world's financial powerhouses afford a project that would stretch the resources of the richest economy? Step forward thin-client specialist NComputing and its cut-down, bloat-free approach to personal computing. The company's X-300 solid state devices allow a standard PC to act as a mini server, powering up to seven thin-client terminals.

ZDNet.co.uk caught up with NComputing's chief executive Stephen A Dukker to find out why Macedonia is betting its children's future on his company's technology, and why thin client is a better answer to bridging the digital divide than a green laptop.

What do you want to achieve with NComputing that other IT manufacturers haven't done already?
We are focused on dramatically increasing the base [of computer users] by reducing the cost. The user market has been stuck for nearly 10 years at 850 million users, which is generally seen as excluding the developing world. Gartner and IDC claim the developing world represents anything from 755 million to 855 million new users, but there are also large numbers in the so-called under-served markets in our own home communities.

In the US, the most crying need for computing suites is in education — as in the UK — where educational institutions are trying to achieve the goal of one computer station per student. In the US we are at the miserable level of one computer station per six or seven students.

If we find a solution to enable this next major wave of users, the implications on existing usage patterns and users are quite profound as well. If we find ways to open up these very large markets — that is primarily economic — we could revolutionise efficiency and usage of computing in the developed world too, as there is nothing in the solution that is unique to the developing world.

The project you are currently engaged with in Macedonia sounds like a massive undertaking, and possibly a record in terms of a thin-client deployment.
As far as we know this is the first country-wide, full education deployment where they made the commitment to equip every single student seat in every single school in Macedonia with a computer workstation, and achieve a one-to-one student-to-computer rating, which is the best in the world.

They will be rolling out 180,000 student seats, of which 100,000 are being done right now over a five-month period. They are going into high schools because those students are the closest to entering the workplace and they want them to be computer-literate. Next year 80,000 additional seats are being rolled out into primary schools.

These 180,000 seats make up 50 percent of the students in the country because they don't have enough classrooms for all the students, so in essence half of the students go to school in the morning from 6am till noon, and the other half go from noon to 6pm. This way, 180,000 seats service nearly 40,000 students.

What's the background to this deal? How can a relatively poor country such as Macedonia justify spending this much money and effort on computing?
With the election of the new prime minister [Nikola Gruevski in August 2006], his particular agenda that got him elected was a major commitment to upgrading the educational infrastructure of the country. If they were to join the ranks of the developed world, they had to be a knowledge- and information-based society.

This is the single largest commitment for funds in the history of the country. They committed €30m (£22m) out of a total country budget of €1.8bn  (£1.3bn) for education. They said that their usual education budget is about €800,000 (£580,000) — so that's a huge step forward. The people in the country agreed to a special tax assessment to help pay for this as they recognised that this was going to be the most important investment they could make.

Most of the former Yugoslavian countries, Bosnia, Croatia, and so on, are going out for education restructuring in 2008, and we hope this is going to be...

...the template for these deployments. It is also interesting to note the Chinese government has also been very active in this area; in fact the company that won this deployment was the Haier Company, which is one of the largest industrial concerns in China.

We are not entirely certain, but we believe there has been some grant allocation by China to help some of these economies pay for this technology.

How much was the Macedonian government able to save by opting for the thin-client route compared to buying regular PCs?
The next lowest bid was twice the cost and was based around thin-client technology from Wyse. The specific metrics are that we have seven users per $350 (£170) PC — which shows how powerful the devices are. [Each PC acts a server powering seven thin NComputing terminals]. Our client devices run the Edge Ubuntu version of the Ubuntu operating system with OpenOffice.

Some developing countries can be resistant to open source as they want to use the software they associate with developed countries — that is, Windows and Office. Was this a problem in Macedonia?
Our product will work just fine under Microsoft Windows as well, but this was purely the decision made by the Macedonian government in terms of wanting to use open-source tools. Microsoft was involved in the programme, proposing using Windows XP Home Edition, but the government decided it wanted to stay with open-source. This was a decision the government made associated with wanting very low-cost infrastructure. [The decision was made] not just for the students to learn technology but, as they transition to the business community, having a very low-cost technology for businesses [is beneficial] as well.

They are looking to jump-start a whole ecosystem in the country around open source, according to their minister for information. Over a five-year period, they want to become an information- and knowledge-based society. This will allow their kids to be competitive with kids from Western Europe, the minister believes.

What percentage saving do you think there might have been with the open-source approach, as opposed to the government opting for Microsoft Windows and Office?
I am guessing in this deployment it is about $600,000 (£290,000) total deployment difference, because I believe Microsoft was offering their per user unlimited potential programme. That is actually quite small considering the entire deployment was about €30m.

The cost of maintenance is quite low, however, and actually relates to the 20,000 PCs that were deployed, rather than the 180,000 thin-client workstations, because the workstations themselves are maintenance-free. They have basically one chip in them, no moving parts — [and] consume one watt, compared to 120W for a PC.

Also, the Macedonian government acknowledged that any technology deployment they made would be obsolete in five years and the costs of upgrading the infrastructure in five years means they would never be able to afford to upgrade PCs every five years. But they can afford to upgrade the 20,000 shared PCs because they don't have to upgrade the attached terminal devices as they will behave like any PC they are attached to.

How does that work with licensing? Does a user organisation of your technology pay per server PC, per thin-client workstation, or another metric?
With our infrastructure, there is no additional charge for the software. When you purchase our terminal devices, it includes our software that includes the multi-user environment. With regards the operating system and the applications, that is unique to the integrator but with Microsoft you would pay per seat. For education they have numerous discounts. In the UK, for example, they have the open education discount programme which can reduce the cost per seat down to about £1 for the operating system and £20 for Office.

So a seat in this example is one user using Office through one of your terminals?
That is correct. This technology was not developed as an educational technology but it just turns out because it is very low cost and does not require certified network engineers to install the software — it's self-configuring. It is extremely easy to deploy, as evidenced by rolling out 100,000 of these devices in five months.

How does the terminal management compare to managing a traditional local area network of PCs? Does it lend itself to the teaching environment in that a teacher can control what appears on a pupil's terminal?
We have a complete set of tools so that if a teacher is sitting in front of a screen, or two screens or three screens or however many, the teacher can observe up to 128 thumbnails of screens, and they don't need any knowledge of the architecture to do this. The teacher just enters the name of the students, and can be observing...

...the students in the physical local classroom. Or, if there was a classroom in London and another one in Liverpool, as long as there was IP connectivity between the server devices, you can manage remote classrooms as well as local ones.

All the tech support can be done remotely, as long as there is an IP connection. This applies to call centres too. If this was being used in a commercial situation, a call-centre supervisor could do the same thing with call-centre staff in terms of monitoring their systems.

Certain teaching skills are very rare in Macedonia — particularly science skills — so this technology allows one teacher in one location to teach multiple classes through real-time, online learning. It is really extremely cool.

Does your thin-client system require high-spec PCs to act as servers, or will it work with older machines?
[In terms of] what class of machine will support our solution, we recommend anything that contains hyperthreading technology or beyond — so a Pentium IV with hyperthreading is a very reasonable machine for this kind of technology.

PCs that are in the neighbourhood of three years to four years old are very adaptable to using the technology.

Which other countries' educational systems have you deployed this technology into?
In June, the Vietnamese education board announced that they had chosen our technology as the standard computing architecture for every school in the country and they intend to deploy it in the next three years. They haven't started deploying it yet because they are waiting on funding.

But it's also being taken up in the US. In North Carolina we actually sold two school districts in that state. Strictly by word of mouth, in six months 17 other districts decided to acquire this technology and deploy 13,000 seats over that period. That has to be one of the biggest viral deployments of technology ever recorded.

Given that this is such a powerful model and so many countries are deploying it, what has held it back until now? We have been told for years that thin client is the future.
If you take a look at how thin-client computing has been deployed up to this point, everything has been focused around enterprise users.

If we look at the components that make this possible — on the software side, to be able to deploy multi-user computing, only three companies can do this. Citrix, a billion dollar company focusing completely on the enterprise; Microsoft itself through Windows Terminal Service — again focused on the enterprise; and, thirdly, VMware. These companies price this technology from $200 (£97) to $400 (£194) per seat just for the multi-user software.

Also, thin clients have traditionally been stripped-down PCs. However, our software infrastructure, which we call the desktop virtualisation infrastructure, has been a 12-year development effort with over 1,000 man years of development time. Compare that to machine virtualisation with VMware, where every user runs their own virtual computer with their own copy of the operating system and their own copy of the applications. Our technology takes a single copy of the application and the operating systems and allows multiple user desktops to be created sharing the same OS and the same physical copies of the applications, which results in incredibly optimised memory performance. VMware's guidelines are 1GB per user. We support seven users per gigabyte.

What we have said is that we are going to take this enterprise technology into an enabling technology, and it has been truly remarkable in terms of the uptake in education and other applications.

It is a very different model to that pushed by Nicholas Negroponte and the OLPC project. He claims that children will only really engage with computers if they have some sense of ownership of the device.
There is about 4,000 to 5,000 years of history and learning associated with teaching children in classrooms. Whether one-to-one usage of PCs versus investing in more teachers and classrooms makes sense — I don't know. There are often very limited economic resources available, and our approach has been to fundamentally reduce to the minimum the cost of the infrastructure. We would rather leave it to the educators and the politicians to decide how best to deploy the technology.

Also, we are a relatively unknown company. We have not had the PR impact that OLPC has, but in the past year we have sold half a million seats — and have two countries basing their whole computing infrastructure on our devices.

The last point I would make about introducing technology where there is not a substantial amount of technology existing is: who does the maintenance, who does the repair, who does the training? There are no IT ecosystems, there are no PC companies in these communities and in the OLPC model it requires the government to take those responsibilities. [It's different from] our model that allows a local IT environment to grow up around it. All our distributions have been done using local partners.

Editorial standards