[Sussex] Ubuntu used in the fight against AIDS

[Geoffrey Teale]

On Sat, 2007-03-17 at 08:22 +0000, Nico Kadel-Garcia wrote:
> That's why I said "it's a sign": I didn't say it's a conclusion. Your 
> responses here have been illuminating: the fact that you paid enough 
> attention to it to respond is even better, and helps allay such concerns.

Well, I try to be and open and honest as I can be.  I was a little
concerned that my cheeky little job ad was backfiring badly on me :-)

> 
> But it's still not clear exactly what you folks do with Linux.
> 
> > ... of course you only have my word for all of that, but there's
> > enough people on this list who've known me long enough to vouch that
> > everything written above is genuine.
> 
> Fair enough. To bring it back to the topic, what *precisely* are you using 
> Linux for and why? Is yours an end-to-end Linux solution, or are you simply 
> using Linux for the backend servers in your data handling? Which 
> distribution? 

Aha.. all good questions.   To explain in detail would take an age -
I've just finished writing a summary of what we do and that ran to 22
pages of A4.   I'll try and summarise the summary :-)

The system was orignally "spun-out" from Oxford university sometime
around 1999.   We make clinical trial appliances - both for server
architecture and for end users. Every single appliance is built on
Ubuntu Linux - some of the server appliances are sometimes deployed as
Xen guests.  Most of the code is written in Python, but there's some C
and bash kicking around, even a fair amount of Emacs LISP.

In a clinical trial carried out using electronic data capture at site
what you'd see is this:

* Off site hosted servers in several physical locations, all talking to
each other and to the laptops, which can be virtually anywhere on the
planet.

* Trial configuration staff working in a CRO or pharma company.  They're
working on virtual appliances running as Xen hosts on a server and
accessed using Sun Secure Global Desktop - these guys use an interface
built in GNU Emacs - though at some point we'll probably end up rewiring
this in Eclipse.

* Data management staff working either on laptop style appliances or on
desktops accessing a server appliance via the Sun Secure Global Desktop
system (which runs anyplace java does).

* Investigators and study nurses capturing data from medical devices,
and/or typing directly into a laptop appliance. 

* Field monitors, read-only/reporting users and some investigators in
large or specialist sites accessing server appliances via XHTML based
interfaces.

What's key to all of the above is this:

* It's a peer to peer system, every machine can cope quite happily
without continous network connections (laptop appliances can be used
anywhere in the world with at least partial, intermittent mobile
coverage).  The appliance will use the best comms it can find, when it
can find them.

* When we say "appliance" we mean it.  Every machine is single-purpose
and as much as possible "just works".  Users never need to worry about
network connections or viruses, equally they don't have to worry about
how you set up a secure network link between, for example, Las Vegas and
Kasane (Botswana).  Short of some cabling and rack mounting in the
server room  nobody has to worry about how you build the appliances,
because the first one builds the rest and configures the network
infrastructure to support everything.  

* Everything can be centrally managed - we can change the very structure
of the clinical trials on machines, distributed across the globe,
without the end user having to even log out their session.

So that's about as short as I can make it, and as a result a little
cryptic - is it any clearer, or are you more confused?   Any questions?


-- 
Geoffrey Teale <tealeg at member.fsf.org>
Free Software Foundation