Sneakernet

Introduction

This article is about implementing a ‘Sneakernet’ for any group of OLPC schools, transporting email/files on USB Flash Drives.

The national-governments that supply the OLPC laptops to schools aim to provide internet-access, if possible. However, in some parts of the world, it’s unavailable or unaffordable at present. To get the full educational benefit of their OLPC Laptops, schools therefore need some other means of communicating with each other.

In places where high-tech alternatives to internet-access by phone-line (such as long distance Wi-Fi) are unavailable or unaffordable, one possible solution is to use a ‘Sneakernet’.

‘Sneakernet’ is a term for transporting information between computers by carrying it from one place to another on a removable medium, such as a USB Flash Drive, CD or DVD. Someone uses their shoes (possibly sneakers) to move data around instead of using the telecommunications network. The term is often used jokingly, but several organizations have successfully implemented Sneakernet ‘networks’, linking villages in developing countries, universities or businesses.

Several people have mentioned ‘Sneakernets’ within the OLPCWiki, such as Motoman. This article draws together Sneakernet ideas from various sources and describes...

How a system could be implemented that meets the requirements of OLPC schools.
The various benefits.
The low cost.
How a manual system could be set-up and run quite easily.
How software might eventually automate the sorting and routing of messages.

The article covers plain Email messages between schools plus file-attachments for distributing eBooks, web-pages, software, course-work, etc.

In the suggested design, volunteer-couriers transport email and files between schools on USB Flash Drives, like cables on a network. It would allow the schools to communicate with each other and, if the network can reach an internet-connected school, communicate by email with the rest of the world.

A typical system might look like this :-

Fig 1 – Example Sneakernet – An outlying school connecting to the internet.

In Fig 1, school ‘A’ can send emails to any other school on the network. It can also pass messages through the other schools to/from an internet-connected school and therefore to any email-address in the world or to a Sneakernet in another region or country.

A Sneakernet is more of a people/organization-project, than a technology-project.

Requirements for an OLPC Sneakernet

The schools need to be consulted about the exact requirements, but they are probably something like...

A simple and understandable system.
Not too much work for the people that run it.
Low-cost.
Available quickly (maybe a simple manual system, with a software-based system later).
A system that any group of schools can set-up and run on their own, using a guide-book, so they can have a Sneakernet without delay (no dependency on an outside organization to run it or waiting your turn for a system).

Overview of an OLPC Sneakernet

In this article, I suggest that groups of OLPC schools without internet-access should set up their own Sneakernet, using volunteer-couriers and USB Flash Drives.

In an ideal world, people would have immediate, real-time access to the internet and the World Wide Web. However, a lot can be achieved with just email. It allows anyone to communicate quickly across the world and send/receive file-attachments. If schools had any type of email system, teachers and children could get all the help and support, information and learning material that they need.

Some organizations have already set up Sneakernet services, taking files from place to place by motorbike or bus, on stacks of CDs/DVDs. A single ‘courier’ visits all the schools in a single day. Although this works well in certain circumstances, it makes the schools dependent on an outside organization to set-up and run the service, which limits how many groups can have it and how soon. The Sneakernet design in this article is different; it’s a de-centralized, collaborative effort among the schools.

An OLPC Sneakernet would consist of a network of road/rail/river/footpath ‘links’ between pairs of neighboring schools, like cables in a wired-network. Each school would be responsible for one link to a neighboring school. Every day/week, a volunteer-courier from the school would transport messages along the link on a USB Flash. It would contain the school’s own messages and any messages passing through that link.

On its way from sender to receiver, each email-file would pass through several links. At each school, the files would be copied from the incoming flash drive to the outbox-directory for one of the outgoing links, according to the destination. The flash drive for each individual link stays in its local area.

Each courier only has to transport messages to/from the next school; they don’t have to travel far from home, making it easier to get volunteers.

They can make use of whatever means of transport is available in their area, such as cycling, walking, bus, train, boat, car, truck, pack-animal, etc.

The benefits of the system come from co-operation between all the schools. The effect of each courier’s work is multiplied because, when a school sends an email, they only carry it along one link at most; the rest of the journey is taken care of by other schools. Also, each courier is carrying messages for many schools, not just their own.

Fig 2 - A USB flash drive, shown with a 24 mm U.S. quarter coin for scale.

See Wikipedia USB Flash Drive.

Fig 2 is Public Domain by Wikipedia author Dori – see Wikipedia Image page for full copyright info.

A Sneakernet could link from two up to hundreds of schools in a region. If possible, it should include a school with internet access, so that any school on the Sneakernet can communicate with internet email-users and with Sneakernets in other regions/countries.

In developing countries, if you think about the current ‘reach’ of the internet, Sneakernets could extend that by another 50 or 100 miles. If a Sneakernet isn’t linked to the internet, a group of schools can still get a lot of benefits from email/file-transfer among themselves.

It’s simple and affordable and could be up and running quickly for a lot of schools.

Choice of Transport Media

Possible transport-media are USB Flash Drive (Memory Stick), Flash Memory Card, CD or DVD.

I believe USB Flash Drive would be the best choice. Files can be transferred to/from them in an instant like another disk drive, they have more robust connectors than memory cards and you don’t have to buy a CD/DVD-Writer.

Low capacity flash drives aren’t expensive; about 3 US dollars for a 128MB drive (in 2007). Even a 128MB flash drive can hold thousands of messages and quite a few file-attachments. Only one flash drive is needed per school. The cost is insignificant compared with all the $100 laptops in a school.

The system can operate with any medium. If a better one comes along in the future, it could be used instead.

Services

Schools could use email for a whole range of services that don’t require real-time communication, including :-

Email (with or without file-attachments).
Voice/video mail messages
File transfer (to access libraries of eBooks, software, photos, video-clips, documents, etc, to upload those items or to upload pages to websites).
Request individual articles, web-pages or copies of small websites by email, from friends or email auto-responders.
Subscribe to email newsletters or web-feeds.
Take part in distance learning by email, such as computer-skills, science, vocational-training, ongoing teacher-training, etc.
Offsite backup of files (children’s work, school paperwork, etc).

Non-real-time communication is sometimes called ‘delayed-communication’ or ‘store-and-forward’.

Volunteer couriers

The people who transport the Flash Drives don’t need to know anything about computers. They could be volunteers from the school or community, or people who are paid to do it.

The volunteers could make special trips to transport the files or they may be people who regularly make the school-to-school/village-to-village trip for other reasons (market-traders, bus-drivers, taxi-drivers, health-visitors, etc).

The availability of volunteers and the needs of the schools would determine the frequency of the service (daily, weekly, etc.).

Analogies to existing services

You can think of the Sneakernet in two ways, as...

A postal service

Letter = One email (with/without file attachments)
Mail-bag = The USB Flash Drive containing multiple messages.
Street Addresses/Houses = The schools
Couriers = Volunteers who take files on USB flash drive from school to school
Delivery-route = The route between 2 schools that a courier takes. Each courier just covers one school-to-school route every day. Collectively, they cover a complete ring/circular-route or linear-route. There can be several linked rings, lines, etc.
Postal-sorting offices = Volunteers at each school (or software), accepting messages to that school then passing the rest on to other schools on the appropriate flash drives for the out-going link.

A wired-network

Node = A school
Branch/cable = Path between 2 neighboring schools
Data packets (constituting an entire file) = Files on a USB Flash Drive
Transport mechanism = Flash drive, plus volunteer to transport it.
Switchers and routers (equipment that bridges ring/lines/stars/etc of the network together and directs the data packets to the right place) = Volunteers (or software) at the school copying each file from an incoming USB Flash Drive to the correct outgoing USB Flash Drive, depending on the route to its destination.

Roles and responsibilities

Users – Children and teachers who use the email service.
Couriers – The volunteers who transport the USB Flash Drives.
Sorting-office workers – Volunteers at each school who direct the incoming files to the right outbox, according to the next link to the destination.

Network topology or shape

The network can be any mixture of different Network Topologies or shapes; one or more lines, rings, stars, buses, trees, meshes, etc, as shown in Fig 3.

Fig 3 – Network Topologies or shapes

Fig 3 is Public Domain, by Wikipedia user Foobaz – see Wikipedia Image page for copyright info.

A linear-chain might be suitable along a road, rail-line, river or valley. For a wide area, the network may be arranged in several rings, branches, star-structures, etc. For villages surrounding a large town, a hub-and-spoke/star-structure might be suitable.

Linking Sneakernets in different regions or countries

Sneakernets in different regions/countries could be linked via the internet email system.

Initially, the work of sending/receiving emails would be done by volunteers at the internet-connected school. They would either send/receive the individual emails themselves or send them all in a single .tar file to another volunteer on the internet for distribution, minimizing the burden on that school and technical knowledge required. Later on, it could be automated by software or a website.

Comparison with existing sneakernet schemes

Of course, some people already pass files around on physical media from time to time, using Flash Drives, CDs, DVDs, etc.

The key features of this scheme are :-

It’s an organized, reliable service, run to a timetable, so that children and teachers know they can always send and receive emails and files every day/week. They don’t have to ask someone to deliver messages/files as a special favor.
Groups of schools can set it up on their own, using a guide-book.
Only the files go from sender to receiver, not the physical media.
Sneakernet doesn’t just move files between two places, it covers a whole network of schools.
A school can send email to a school with internet-access and hence, to other regions, countries and Sneakernets.
There is one courier per link, not one for the whole network.
Each courier only has to move the files a short distance, to the next school. Other schools take them the rest of the way.

Design options

There are two possible types of Sneakernet; either sending the physical-media all the way from sender to recipient (as in some existing schemes) or sending just the files (as in this scheme). Different groups of schools will have their own opinions about which option is best for their circumstances. Each region’s Sneakernet doesn’t have to use the same system.

For widely-separated schools in rural areas or large numbers of schools, sending files is probably the best option. The suggested design assumes this is the most common situation.

For schools that are close-together in urban areas, taking a single flash-drive around all the schools in a single day might be the best option.

Alternative file transport methods

OLPC laptop

One alternative to USB flash drives is to transport files between two schools on an OLPC laptop, connecting wirelessly to each school’s Wi-Fi Mesh. This assumes a laptop is allowed to access another school’s mesh-network without reconfiguring permissions each time. This design-aspect of the laptop hasn’t been decided at present (July 2007). See Question on Accessing another school's Mesh Network.

Using an OLPC Laptop has the advantage that everything could be automated in software, so all the courier has to do to collect/deliver messages is to come within range of the Mesh Network at each school. The software does the transfer.

If software is written for the server-computer, to sort and route messages, there would be no need for ‘postal sorting-office’ volunteers.

Some disadvantages of using a laptop to transport messages are :-

Software would have to be specially written.
A laptop is more weight to carry than a flash drive.
You may not want to let non-school volunteers carry the laptop between schools in case it is lost or damaged. This could limit the number of volunteers.

Letter or parcel post

If there is an affordable postal- service in the area, it could be used to send flash-drives or flash-memory cards between the schools or straight to someone with an internet connection. The Digital StudyHall Postmanet, for example, uses postal-distribution for video DVDs of school lessons.

Telephone file transfer

In areas with a telephone system, but no Internet Service Provider, one of the pre-internet systems for ‘email-transfer by telephone’ could be used for some/all of the Sneakernet links or to transfer files from the network to a computer with access to the internet email system.

The Wizzy Digital Courier project in South Africa uses UUCP Unix to Unix CoPy. There is also the FidoNet Bulletin Board System. These make use of night-time cheap-rate calls. The receiving computer has software waiting for a call on an auto-answer modem.

Another approach is to arrange a time for two computer-users to transfer files using a terminal program.

If off-the-shelf software is used for these methods, some education in communications and a step-by-step guide would be needed. If possible, transfers would be automated by specially-written software.

High-Tech Software or Low-Tech Manual System

Manual system

A low-tech manual version of Sneakernet could be started straight away, where the sorting and routing of the files is done at each school by volunteers. They just need to look at the list of incoming files, consult routing-tables and diagrams, then drag and drop groups of files to the appropriate outbox directory and outgoing flash drive. All it requires is a procedure that they all agree on, published in an ‘Operator/User Guide’. No special software would be required.

Some method of routing messages to their recipient would be required, such as this file-naming scheme:-

<Destination School No.> <Recipient No.> <Source School No.> <Sender No.> <Message No.>.txt/.tar

If you are school 0037, then any incoming files named 0037*.* are for you, because filenames begin with the destination school-number. The rest of the files are placed in outboxes for other links, to go to other schools.

Putting the <Destination School No.> first means that all messages for each destination school are kept together when you view an alphabetically-sorted list of files. This makes it easier to select a group of files and copy them to another outbox directory or flash-drive.

For example :-

Dest Src  Rec  Msg
0001 0001 1234 5678.* } For school 0001
0001 0002 1234 5678.* }        “
0037 0001 1234 5678.* } For school 0037
0037 0003 1234 5678.* }        “

When selecting the output-route/flash-drive, the ‘volunteer-sorter’ would refer to a diagram of the network or a table of outgoing-links and destination school numbers.

Emails would be written as .txt files. Private emails or those with file-attachments would be put into a .tar container-file.

External emails would need email-address information, for the benefit of the internet email-system and the person at the internet-connected school who sends the emails. The information could be in the filename, in a separate file or some other means.

A manual Sneakernet system would be quick to develop, but needs some file-sorting and routing work each day.

Operating a manual system for a while might be a good way of finding out the exact requirements for a software-based system.

Scripts to assist a manual system

Command-line scripts might be used for file-sorting/routing. For example, if a school had links to three others, then files coming in on one link could go out on two others. A script such as this would copy the incoming message files to the right outbox, according to the destination school-number :-

cp 0000*.* /outbox2
cp 0001*.* /outbox3
cp 0002*.* /outbox1
...
cp 0037*.* /thisschoolsinbox
...
cp <Max>*.* /outbox2

The script has one line for each possible destination school-number and its corresponding outbox directory. Each school would have its own script reflecting its School-Number, position in the network and incoming/outgoing links.

At the destination-school, another script could copy emails to each child’s inbox, according to the Recipient No., so at school 0037, it would be :-

cp 0037 0000*.* /child0000inbox
cp 0037 0001*.* /child0001inbox
etc.

Automated software based system

In an automated software-based system, a computer program would do the same message-sorting job as the script. It would read files from the incoming flash drive and copy them to the correct outbox or outgoing flash drive. Software would also transfer emails from a standard email client program to the school’s inbox and outbox.

The basic ‘unit of transmission’ might be the user-file (email/document/program-file, etc), as in the manual system. Alternatively, small ‘data-packet files’ could be used to emulate the internet (TCP/IP) and provide a better platform on which to build many different services. Data-packets would interface more easily with existing software, such as email-servers and web-servers. An automated system should ideally be based on internet standards, but adapted for delayed networks and time-outs measured in days. Software for delayed networks may already exist.

For data-packet files, the file naming system might be based on the Network IP Address of the recipient’s school’s server, such as :-

<Destination IP> <Source IP> <Packet No.>.tar/.txt

Email User-Interface

Manual system interface

Users who wish to send a message would write them as a text file, either by using a text editor or by using an email client program and saving each message to a file. Messages from all the laptops would be transferred to the school server.

To hide the technicalities from the users, a single volunteer at the sending school would ensure that each message has an appropriate filename (Destination School No., etc) and put private messages and any file-attachments into a .tar container-file. The process would be in the operator’s guide and could be manual or use scripts.

Users would follow a standard template for emails, such as :-

From: John Smith [jsmith@olpcschool0001.edu]
Sent: 26 July 2007 20:06
To: Fred Jones [fjones@olpcschool0002.edu]
Subject: Invitation to school sports day
<Message body>

Automated software system interface

In the automated version of Sneakernet, children would use a standard email client program on their laptops to read/write emails and software would interface it to the school’s inbox/outbox directory on the server (perhaps via an open-source mail-server program). This would be convenient and also allow children to experience e-mail in the same way as everyone else.

When the courier is ready, software would copy the email to/from the USB Flash Drive. It might even be set up to run automatically, when a flash-drive is inserted.

An installation program would take care of setting it up. Nobody at the school would have to get involved with the technicalities of how the software works.

Who could use it

Sneakernet could be used by any groups of schools that are reasonably close together, perhaps 1 hour to half a day’s traveling time between pairs of schools. They could be in cities, suburbs, rural-areas or mountainous-areas.

They would be either :-

A group of schools with a connection to the internet via one of the schools, or
An isolated group of schools, just benefiting from email and file-transfer between themselves.

If a school is 100 miles from anywhere with internet access, that doesn’t matter. The only important distance is the distance to the nearest school on the network. If the school’s courier can get there and back in a reasonable time each day/week, then they can ‘connect’ to the network and send/receive emails over long distances, to other schools on the Sneakernet and (via a school with internet-access) to the rest of the world.

Very isolated schools might pass a flash-drive from person-to-person/village-to-village, to reach the next school on the network.

No doubt friends and family and people in the community would want to send/receive the occasional email.

It would be best to start small, with 2 or 3 schools, get the system working for a few weeks, then add more schools to the network one at a time.

A good place to try it out would be a chain of schools with an internet-connected school at one end, perhaps along a linear route, such as a main road, river, rail-line or bus-route.

The next stage would be a more-complex network of schools, with multiple rings and branches, perhaps on the outskirts of a large city.

The schools that use a Sneakernet would change each year as the internet extends further and further. It would be useful somewhere for years to come.

Privacy of messages

Because messages pass through many people’s hands, users may want to password-protect them for privacy, by keeping the message body and attachments together in a password-protected .tar/.gz container-file.

Implementation

A manual system just requires a written operator/user-guide (plus a few scripts if desired).

For a software-based system, it’s just like any other software-project. A design-team would publish a consultation-document, talk to users and agree the precise requirements.

They would then produce various outputs, such as :-

A detailed design.
Software
Test Specifications
Test results
A setup/installation guide
A user-guide and training material for children and teachers.
Instructions and training material for the volunteer couriers.
A software-maintenance guide.
A website or e-mail mailing-list for distributing software and documentation updates.
Forums for users and developers.

These would be made available as open-source software/documents for anyone to use and adapt to local conditions. They would also be given to new groups of schools when they get their OLPC laptops.

Optimization

When a simple system is up and running, further development-work could be done to minimize delays, files-sizes and storage requirements.

Link to the internet email system

Each Sneakernet would ideally include a school with an internet-connection, so that all the schools on the network can communicate by email with other internet-connected schools, internet users and Sneakernets in other regions/countries.

The cost of internet-access would be shared. If no school has internet-access, one may be near a small-business/home-user/internet-cafe. If one of the schools has a postal-service, then a USB Flash Drive could be mailed to someone with internet-access.

If internet-access is unavailable, a group of schools could still get a lot of educational and social benefits by operating a Sneakernet service between themselves.

Additional services

Once an email system has been running for a while, additional, automated services could be set up to make more use of it, trying to replicate some web-services using email on a delayed network, such as...

Web-access by email

Requesting files from an Email Auto-responder, to get web-pages, copies of small websites, web-directories, search-engine results, eBooks, Software, learning-material, etc.

Public email auto-responders like this already exist, but they are often overloaded and not working. One restricted to OLPC schools would be better.

The ability to obtain web-pages would give children and teachers much greater access to information and some limited hands-on experience of the World Wide Web. The service could be restricted to web-pages from a private OLPC internet for a particular country, to ensure content is suitable for children.

Mirror-sites

A service that automatically maintains mirror web-sites at a school, for fast-access. These could be copies of websites from the internet, a private OLPC network, schools on the same Sneakernet or other Sneakernets. People would use the service to subscribe/unsubscribe to particular sites.

If the hypertext links worked between a collection of mirror-sites, it would give children more experience of browsing the World Wide Web. A large collection of websites could be accumulated slowly over several weeks and months.

Designing services for delayed networks is a whole subject in itself.

Manual version of services

Until software can be written, these services could be provided manually, by sending requests to a number of friends/colleagues/volunteers with internet-access.

Advantages of a Sneakernet

Most importantly – A Sneakernet is much better than nothing (no email/files at all).
Any school can start one. There’s no need to wait for an outside organization to run the service or provide transportation.
The courier from each school only needs to travel a short distance, to the next school.
Cheap equipment (just one USB Flash Drive per school).
One shared internet fee (or none, for an isolated group of schools).
Simple and understandable for the users and operators.
No electric power requirement during file-transport.
It gets people interested in faster communication technologies.
Low financial risk (if the project stops, only a few dollars have been spent per school).
It can operate with little training and support (some can be via the sneakernet itself).
It doesn’t leave schools with a lot of high-tech hardware to maintain, like some communication systems.
Unlike radio and Wi-Fi systems, there are no problems with telecommunication regulations, monopolies or licenses.
It can start simply and add more features later.
It can be easily modified for local conditions.
It lets schools help themselves.
Local people can access government/business-services and information remotely, saving travel-time and expense.

Disadvantages

Its not full real-time access to the internet
It doesn’t give children the full experience of the World Wide Web
Time-delays of many days to send a message and get a reply
It requires volunteers prepared to transport the flash drives, etc.
Transport may be difficult in bad weather, monsoons and difficult terrain.

Cost

Each school requires 1 USB Flash Drive (3 US dollars in 2007).
The software would be free and open-source.
There are no on-going monthly costs for an isolated system.
If there’s an internet-link the cost could be shared by a large group of schools.
Schools may want to pay someone a small amount for transporting the flash drive or pay to send flash-drives/flash-cards by post.
Costs can be controlled by having a daily, weekly or even monthly service and limiting the number of external internet emails.

Benefits

These are much the same as having internet-based email :-

Schools can use email/file-transfer and obtain whole libraries full of eBooks and teaching materials.
Teachers can communicate with colleagues at other schools for help and support and with local education-boards.
Children and teachers can communicate with other schools for educational and social reasons. For example, children can have pen-pals at other schools, send letters and photos and co-operate on projects.
Teachers can use email to arrange sporting fixtures, etc, with the other schools.
Children can use distance-learning by email for a wider range of subjects.
The wider community could get information and keep in touch with friends and family elsewhere, and access businesses, local-government, health services, etc.
The couriers could deliver other things, such as charged-up batteries for schools.
Teachers and children could publish locally-produced content on websites, so that other schools can access it (web-pages, eBooks, eZines, stories, artwork, music, video, newsletters, blogs, teaching materials, etc). This is an important aspect of the OLPC project. It means that people aren’t the passive recipient of content produced elsewhere. They can create items reflecting the local language, culture and interests and publish it for the benefit of other schools.

Who would develop Sneakernet?

Any computer-organization that works with schools in a developing country.

A manual system could be developed quite quickly. The schools just need to agree on how they want to operate it. An operating-procedure/user-guide would then be written. Some command-line scripts could also be written, to sort, route and copy messages plus instructions on how to customize them. These would then be published and used/customized by any group of schools.

A software-based system might make a good university/spare-time open-source project.

Most of the development-work could be done in the lab before testing it in the field with a group of schools. If the Sneakernet which is used for testing included an internet-connected school, then people in various countries could co-operate on the project.

Further development

The system user-guide, software and training material would be made open-source, published on the internet for anyone that wants to use it, and maintained and developed by the open-source community and the schools themselves.

Sneakernet projects using the OLPC laptop and conventional computers might use similar software, so all the projects would benefit from any developments.

The final design of the system

The final design doesn’t have to be as described here; it could include the developer’s own ideas, ideas from the schools who are going to use it, features of existing systems, etc. The important thing is that any email system is ‘much better than nothing’. Schools would get a lot of benefits, even from a really simple system. They just need the first version as soon as possible. It can be improved later.

If anyone is interested in developing an OLPC Sneakernet system, please post your comments on the OLPCWiki discussion page for this article or contact me at [ricardoolpc@yahoo.co.uk].

--Ricardo 02:05, 11 August 2007 (EDT)