Feature - FOOTWAYS takes its first steps

Footways started quite literally in the back of someone’s garage. Image courtesy Igor Dubus

Google, Hewlett Packard and Apple all had humble beginnings in the back of someone’s garage. Could the same be true in the back of a garage in Orleans, France?

“I had never seen a router/switch in my life, I had to get into VPN, network, perl scripting. I also had problems with electricity supply and consumption — this was my home, not a dedicated IT room,” says Igor Dubus.

This was only the beginning. Having built a 96-node cluster in his garage to run computer models as the coordinator of the FOOTPRINT project (iSGTW ran an article about FOOTPRINT earlier this year,) Dubus launched his own start-up company, called FOOTWAYS. His goal is to develop this “garage-cluster” into a 12,000-node high performance computing center dedicated to pesticide modeling.

FOOTPRINT, an EU project, seeks to minimize water contamination caused by pesticides — a matter of growing concern in a part of the globe where an estimated 40% of all surface water has already been contaminated. The project classifies agricultural land into about 100,000 “agro-environmental scenarios,” or simulations of different combinations of soil type, crop, climate and other variables. FOOTPRINT then uses computer models to calculate how each agricultural environment will respond to the introduction of as many as 100 different types of pesticide, allowing the user to predict whether using a particular pesticide in a particular place will cause water contamination.

This technique is called “pre-modeling,” because the processing work is all done before the user asks for it. The results form the basis of a giant database, which can be accessed using FOOTPRINT software tools. Because pre-modeling requires anticipating what scenarios different users will want to run, several hundred-million runs were required, demanding some substantial computing resources.

The FOOTPRINT method of pre-modeling agricultural scenarios had the advantage that users can quickly access the results of complex simulations. But what about scenarios that are not covered by the database? Some users found the scope of FOOTPRINT too restrictive and, because of its nature, the modeling relies on some simplifications. 

A step up in size and complexity. Image courtesy Footways.

From FOOTPRINT to FOOTWAYS

Spotting an opportunity, and having already built the “garage-cluster,” Dubus established the company FOOTWAYS.

Using around €13,000 of his own money to buy 24 tailor-made machines, cabling, a switch and a router, Dubus cobbled together a personal 96-node computer cluster in his garage by himself to run FOOTPRINT computer models. “There were lots of difficulties setting it up since I’m not an IT person!” commented Dubus.

His aim is to allow paying users to run real-time models tailored to their requirements on a distributed computer system.

“The idea is to give users access to a web portal, ‘FOOTWAYS-PRO,’ where they can specify what they are interested in, in terms of modeling and risk assessment,” said Dubus. “FOOTWAYS intends to provide long–term support for the free-to-download FOOTPRINT tools, as I believe that the tools, if used by farmers across the EU, could significantly contribute to the reduction of the contamination of water resources by pesticides.”

FOOTWAYS-PRO is more flexible than the FOOTPRINT system. Not only can the models be custom-designed for each user, but the input and output operations can be customized. For example, a farmer will most likely want results in the form of a simple message or map, whereas a risk-assessment specialist will want to review the specific details of each modeling run.

“It allows risk assessments for pesticides to be undertaken at a level which has never been achieved before, in terms of scientific quality, spatial coverage and flexibility. This is really a next-generation tool” enthused Dubus.

Dubus’ “garage-cluster” now resides in an office and forms the basis of the FOOTWAYS computer network. A dedicated distributed computing administrator was recently hired to start setting up the rest of the high-performance computing network. The aim is to establish 1000 nodes by mid-2010, 6000 nodes by mid-2011 and ultimately 12,000 nodes by mid-2012.

As with all start-up companies, there has been a constant stream of challenges — from financial and organizational issues (such as finding a suitable building), to the technical problems with setting up a computer center. Most importantly, FOOTWAYS needs to establish a strong client base.

“The FOOTWAYS start-up company was started less than three months ago and I think we can realistically consider that very good progress has been made since then. I expect the first contracts with clients to be signed by the end of the year,” Dubus told us. “Putting all this together has been, and remains, an exceptionally interesting experience and we are all very enthusiastic about it. To infinity and beyond!”

—Seth Bell, iSGTW