Thanks for the very informative answer and suggestions. In response to one of your questions about why I think I need such a high end graphics card, I need to do "real time" (ideally at 30fps) processing of 3D data. My current prototype (which isn't ingress protected in any way) runs at around 10-15fps, and with the GTX970, performance bottlenecks are no longer on the image processing side of things (uploading the images to the GPU now takes more time than the calculations), which they were using an older GTX650ti. Unfortunately, moving the data elsewhere for processing isn't an option - there is no guarantee of decent connectivity and the amount of data will be at least 25GB a day. It doesn't need to have a permanent display though, which makes life somewhat easier!

I can certainly make the box larger as you, and other slashdotters, suggest. In fact I've not "been" painted into a corner, I've painted myself into a corner and am always very happy to look for alternatives as you suggest. I'm a big fan of the keep-it-simple mentality! Ideally, an off the shelf solution would be perfect, but having talked to a few of the embedded pc manufacturers as well as endless googling, it seems that I'm going to have to design/build the solution myself. I very much appreciate your answer and clearly you have a thorough understanding of the types of issues I'm facing, and in a way I'm glad that I haven't missed something blindingly obvious!

I'm also realistic in that I don't expect the first prototype to work without hiccups, but if I can incorporate as many of the suggestions as possible in the design, then I can hopefully keep the failure/damage/rebuild rate as low as possible.

Thanks again for your post.

... and very timely of The Register to bring it up: http://www.reghardware.com/2011/08/18/heroes_of_tech_david_may/

I take it then that you've never actually been to Mongolia?

YMMV I guess. What was wrong with it?

It's always worked a charm for me. I wouldn't post a referral link to a broken product! Well not purposefully anyway.

From https://spideroak.com/referral/

Is this not correct then?

Well spideroak allows syncing of different folders on different machines, so in that respect it's more like dropbox on steroids than mozy.

Additionally Spideroak has standard backup features that numerous providers give (including mozy) but gives you up to 50GB storage free.

Why not use Spideroak instead of dropbox. Spideroak have a zero-knowledge privacy policy. I'd say it's not quite as polished a product as dropbox, but everything is encrypted before it leaves my computer (come on spideroak open source your client so we can check!) and stored encrypted, so NO ONE can read it. I have access to files from android to. (I am not affiliated with Spideroak in away way.) Join via this link and we both get an extra 1GB (I believe you start with 2GB free): https://spideroak.com/signup/referral/dd998cb68d2fba5eb916a000411c2263/

... allows potentially very long passwords, are easy to remember and you can always swap out vowels for digits or symbols. If the site doesn't permit spaces then swap them out for asterisks/underlines/a different character/omit the space

http://www.baekdal.com/tips/password-security-usability?

Apparently

http://yro.slashdot.org/story/11/04/14/1718258/European-Court-of-Justice-To-Outlaw-Net-Filtering#

I though this implies that NO ISP can be forced to filter the internet. Surely parts of the DEA require exactly that (IP blocking for example).

'...what do you think they'd do if they thought for a second their tax was going to help write computer video games?'

I think they would be angry.

Let's hope they don't think that as it would be incorrect.

They are not receiving public money, they are (were) just not being taxed on all their earnings. There's a difference here which you may have missed.

No I haven't RTFA'd but is this to do with the East Anglia University mail leak which is turn just shows that in order to make pretty graphs to justify your existence for large funding boards, you have to merge data from different data sets. IIRC there is anomalous tree ring data from the 1960s onwards, so it's more accurate to use actual temperature measurements but these don't match up perfectly with the pre 1960 tree ring data, so a small amount of "fudging" is required. Remember this was to create a graph of temperature over 4000 years for a high level overview - it is not part of the actual analysis.

That's right - put your heads back in the sand. The world is not getting warmer, and at least if it is, God would make sure that it didn't get too hot so as not to harm us. Damn those pesky scientists.

Anybody who is familiar with Red Dwarf will get this "genius pairing of references", let alone those with a modicum of education.