Harsh Computing: Design and implementation considerations in software and hardware systems for remote and rural deployments

From DevSummit
Jump to: navigation, search

Facilitated by Jeff Wishnie, Inveneo

Inveneo, a San Francisco based 501c3 non-profit, implements and deploys communication and computing technology throughout remote rural areas in sub-Saharan Africa. Inveneo's CTO, Jeff Wishnie, will describe some of the considerations that go into the design, implementation, and deployment of hardware, software, and network systems that must operate without administrative intervention, in the absence of tech savvy users, and in hot, humid, dusty, and otherwise harsh conditions.

Harsh Computing: Design and implementation considerations in software and hardware systems for remote and rural deployments/notes

session notes: harsh computing -highlight tech problems we they have and how to get around them, open it up to ideas from group. -comm tech makes it more possible to provide services like food and water to developing world -health care information systems, econ development, connecting to the rest of the world. giving people a perspective outside of village.

-challenges: power is #1 problem, either shortage, or dirty power that will kill equipment. heat and dust, 110-120 degrees. critters, bees, takes equipment down.goats like to chew things that are plastic. kids like to break things because they don't know what they are. people who don't know what they are doing.

design goals for sustainability: suable, affordable, supportable, resistant, resilient, low-power we take for granted that computers just work in developed world, but in developing world, just fixing computers is a large part of the tech experience. stateless systems

design considerations: don't make users become experts. don't expect users to follow directions. design problems out of the system up front. brittleness is bad. make things repairable,

group discussion: internet cafe and thumbdrives are a huge source of viruses virus protection infrastructure is non-existant in the developing world, which is a huge problem for choosing an operating system. lack of power infrastructure is a factor in cutting time running systems, which includes cutting out scanning for viruses. a windows system take roughly 3-4 months to become completely useless because of viruses. use a system (like a live CD boot) that doesn't write anything to the local disk. everytime the machine is rebooted, the system is back online in it's original state. the main focus is on keeping the system booting the next day, it might still be a carrier of viruses though.

computers are 8w and screen is 12W, total 20W at peak operating. key is low power, which makes solar panels viable multiple computers. even doable for generators and buying fuel. servers: 25W peak draw, no fan, operates at 50 degrees C! all data is stored on these, so they have RAID1 drives. connectivity issues: costly. satellite is slow and expensive. 60k connection is $400-500 in sub-saharan africa. all net connections out of africa is satellite. fiber is going into use in africa. Also DSL. projects typically use ICT links to extend the best connections from the central offices out to rural areas.

Radio mobile project is a mapping software that shows topography of land between two radio links and see if there is clear line of sight or not.

The developing world is caring about power consumption which is producing power saving chips and other hardware.