Mist Vs Cloud #1 – The DOT (Disks On a Train) Network and biz ideas for the Postal Department

I happen to live in Bhopal, Madhya Pradesh, India, which sees a lot of clouds of the rainy variety (at least in the monsoons) but is pretty much an arid desert when it comes to the clouds that rain data, i.e. cloud computing services like Gmail, Facebook and Youtube.

Except for oasis like pockets, most of Bhopal has at best “2G” connectivity which, for those who were born into and live in 4G heaven, feels like an old 56k modem from the users standpoint.


While this may seem counter-intuitive to metro Indians many of whom are now getting ready to be born into 4G+ heaven, people who live in Bhopal and cities /towns/villages smaller than Bhopal would probably agree vehemently … except that this page would take too long to load where they are, and lets face it, there are other things on the Internet that are far more interesting to wait for. They probably wouldn’t have the patience to read it. Besides, even if they did get online long enough to browse, Gmail and other such “cloud” services would be chewing away the bandwidth trying to load the latest “smooth transition” or “tabbed inbox” or “predictive text” and they would quickly close all other windows/tabs on the browser to “make room on the pipe for Google”.

As a result of this scenario, most of us at Mojolab have very little faith in the “cloud”(s). Information still remains as important as water to all of us, so we’ve come up with our own alternative to the clouds to serve our hydration needs. We call it “The Mist”.

Unlike cloud computing, mist computing uses local resources, like your buddy’s cellphone data card and your neighbors’ wireless connection, and doesn’t spread farther than the local resources allow, e.g. the network stops when the first neighbor refuses to share his wifi.

While clouds are up there, the mist is down here. When you need to get soaked, you take a walk into the mist, you don’t do a raindance waiting for the clouds to dribble.

What this looks like in less poetic and more technical terms is that instead of relying on a distant “World Wide Web” you start seeking to fulfill your information needs with local peer to peer sharing before blindly downloading or uploading the data to the Internet over a third party connection.

The most unholy waste of bandwidth I have ever seen (and indulged in) is two people in the same room, with two different mobile Internet connections watching the same video on their different devices….in the process paying the same provider twice for the same data. It would be far more efficient to download the video once and then share it with everyone in the room via bluetooth/wifi.

So we don’t stream any more…we “edge cache” and share locally. We don’t share URLs, we transfer files. When we have people in the same location, we use a projector ad a round table instead of a shared document. And we’re trying to build more and more tools that can help us get more productive with our local peers instead of trying to do more and more data intensive things remotely, simply because “that’s where the server is”.

We’re also taking a fresh look at long distance data transfer. The other day I found myself with some idle time on the Internet and did this little thought experiment around what it takes to send 2TB (2 terrabytes, which is roughly 1500 1 GB movie files or 175,000 pictures in PNG format taken from a 28 megapixel camera), from Bhopal to Delhi in the shortest possible time for the lowest price possible.

If I use a Tata Docomo 3G connection (which is the prevalent and most functional network in the Bhopal Area, as verified by OpenSignalMap above), I am paying roughly INR 450 for 3.5GB @ 7.2Mbps (expected 3G speed), according to the image below from tatadocomo.com:


The best speed I have ever seen on a 3G dongle in Bhopal is 2Mbps, but we’ll be generous and say that there are spots where it may be as much as 4Mbps.

Now say I put myself on the Bhopal Railway Station with a solar powered laptop past its amortization period (so effectively free from power and equipment costs) and my friend with a similar setup at New Delhi Railway Station.

I connect my 2TB hard drive to my computer, fire up a peer to peer file sharing client, hook up to my buddy and start transmitting.

By the time the transfer completes, we would have spent about 2TB/4Mbps= ~48 days (each, i.e. a total of 96 man days) parked at a railway station, barring connection drops, bad weather and suspicious cops.

INR 450 for 3.5 GB comes to roughly 128 rupees per GB. So over our 48 day stay, we would also have spent 2048 X 128 = 262,144 (!!!) rupees on 3G connectivity.


By contrast, if my friend hopped on that train (lets say a Shatabdi, one of the higher priced, fast commute trains between New Delhi and Bhopal that leaves Delhi in the morning and returns there by late evening), he would be in Bhopal some time in the afternoon the same day (in air conditioned comfort with a desk, a chair and a shared power socket with food included, so he isn’t adding any more expenses over the ticket either), could copy the 2TB to his computer while the train gets cleaned at Bhopal station (since I am waiting there, remember) and be back in Delhi before midnight (or shortly after if its foggy).

Lets say it gets really foggy and he gets back a full 24 hours later, which is about 1.5 times the expected time.

I would have saved my entire 24 hours, but lets discount that too. Lets say I really want to make sure the data gets across so I keep sitting at the station temple praying for my friends safe return home and we spend a total of 24 hours each, just to be generous give the telcos a nice handicap.

The Shatabdi ticket is about INR 1000 (1037 to be exact, see image below).


So for the cost of 2000 rupees and one day, we’ve gotten across 2TB from Bhopal to New Delhi. This is less than 1/48th of the time and less than 1/125th the money it would take to do it online.

Not to mention that my friend could have carried back a full rucksack of 1TB hard drives, since he’s allowed to bring 40 kilos of luggage on board the train…just to make the proportions even more ridiculous. And if we needed to do this regularly, we could probably save on the tickets by befriending the Shatabdi attendants, who are remarkably friendlier than the folks who answer customer care at my telco.

Whats even more interesting is that even if the provider was to make the cost of the data transfer nil, the time involved would still make the trip worthwhile. In fact to provide an “at par” solution, the telco would have to give me a 194 Mbps connection for less than 1 rupee per GB. With telcos already crying foul about dropping ARPUs and razor thin margins, I don’t see that happening anytime soon.

And the clincher is that these economies exist between any two points at least in India, since it’s always possible to get someone from point A to point B within the country (and to a lot of places outside as well) for less than 250,000 rupees and 48 days.

All it takes is getting past the illusion that the cloud is “free and instant”. It’s neither.

So I find myself constantly amazed that people in India (including myself) remain willing to pay over a hundred rupees a GB for data. India is truly a magical and hypnotic land! But I suppose in a world where people starve despite humanity having known how to store food since before the time of the Pharaohs, it’s a little petulant to cry about data being too expensive. 

Cynicism aside, what I would love to see is the Indian Postal System pick this idea up and start transporting hard drives instead of letters. With the routes and supply chains already in place they could literally auction off the “bandwidth”, which would be best used by commercial interests with reliable large data transfer needs.

This in turn would leave the airwaves free for the rest of us, reduce the need for high power, cancer causing, bird disturbing cellphones and towers and more than enough data coming and going to keep everyone well watered…without the need for a cloud. However, the Indian Postal Service seems to have quit without a fight in the face of the e(G)mail, so we may end up having to reinvent the wheel on this one!