Monthly Archives: May 2006

Geograph to use Ubuntu?

I had been planning to use Debian Sarge on the new Geograph servers, but the buzz around tomorrow’s new release of Ubuntu is hard to ignore. I’d previously dismissed Ubuntu for server use assuming it was too focussed on the desktop, and while they are doing admirable work in that area, it seems they aren’t slouching when it comes to servers either.

Tomorrow’s release of “Ubuntu 6.06 LTS” includes a Server Edition which will be supported for 5 years, with a “certified” LAMP application stack. The certification is primarily to do with MySQL, which I’ve found to be problematic to build from source in the past, so this sounds good to me!

I’ll give it a whirl when it’s released, but if it delivers a low-maintenance LAMP server supported for 5 years, it will hard to resist using it for Geograph. I’ve no doubt they’ve got a great set of packages for LAMP, but one thing in particularly I’ll be looking at is installation of the packages necessary to support LVS based clustering.

More Tea Vicar?

Well, the important question of what to name the Geograph servers has been answered.

  • Tea – the load balancer, as everything should start with a nice cup of tea
  • Toast, Scone and Crumpet – web servers, keeping our site sustained!
  • Jam – NAS / database, as toast, scones and crumpets are naught without jam

See, there’s method in it.

While I’m waiting for the new kit to arrive, I’m building some VMWare virtual servers of the new setup, which should help anyone developing for Geograph since it allows them to run an identical setup without needing to know much about Linux configuration, and will also allow us to trial new ideas without messing up the “real” boxes.

O’Reilly turned evil?

I’m hoping there’s been a few crossed wires and misunderstandings surrounding O’Reilly’s lawyers issuing a Cease and Desist against use of the term Web 2.0. O’Reilly are the “Google” of publishing, surely they can do wrong? Geeks worldwide take a pride in their bulging shelves of O’Reilly tomes bedecked in animal engraved covers. There’s a few people venting their frustration in comments on Tim O’Reilly’s blog, but surely some sort of official clarification must be coming!

Say it ain’t so!

(Edit #1) here is a first response, but it’s not going down well.

(Edit #2) …and a second one! No-one denies they have to make a buck, but is no one at O’Reilly actually grasping the irony of this situation?

(Edit #3) …finally Tim O’Reilly responds. Huzzah.

The Naming of Servers

I’ve done all the hard work, the servers are specified, they are ordered and now we must wait for delivery. Now is the time for that most important of tasks.

What to call them?

I think it’s a trait of all geeks to want to give their equipment cool or amusing names. At one of my first jobs in Newcastle we named our equipment after our favourite drinking establishments. I remember I had Trillian and Baltic. All of my own gear has Star Wars names. I’m writing this on Geonosis but it will pass through the frozen wastes of Hoth on its way out into the big wide world. My family photo album is on Naboo.

It’s not sad, just human nature, honestly.

Anyway, we’ve got 5 servers for Geograph, and I’m appealling to the happy bunch who inhabit the site forums to come up with the names.

I’m thinking rodent based names will be popular given our insistence the current server is powered entirely by hamsters….

Geograph’s New Hardware – from a Skateboard to an Airbus A380

Finally got all the quotes in for Geograph’s brand spanking new hardware platform and we’re ready to order. It’s all rather exciting.

What Geograph currently runs on…

  • 700MHz Celeron server with 512MB RAM and 70GB hard disc space

…and what it will be running on

  • 3 x application servers with dual 3Ghz Xeon CPUs and 4GB RAM
  • 2TB storage server with dual 3Ghz Xeon CPUs and 4GB RAM
  • single CPU load balancer / cache server
  • full remote IP power switching and KVM for all hosts

It’s hard to compare the two setups, but I’d say the new configuration is around 40-50 times more powerful. It’s like trading up from a skateboard to an Airbus A380.

While this will provide a much needed boost in site performance for our present site contributors, some of that horsepower is needed to ensure we can cope with the increased traffic from educational use.

I’ll write more about the process of configuring the hardware over the next weeks, but for now, that’s enough to whet your appetites!

Ordnance Survey 2.0 – Bring it on!

Ed Parsons blogged his personal thoughs on “Ordnance Survey 2.0″ – pretty frank stuff, given he is the CTO of Ordnance Survey.

He writes about allowing free access to OS data and letting “…users discover the value in OS data by actually deploying it, and if the value is there, they will pay for it later.”

I’ve no idea what the commercial ramifications of such a move would be for the OS, but I’m pretty excited by what it means for small companies and back-bedroom innovators. Trial your ideas quickly and without risk. If they turn out commercial, great, everyone wins. If they don’t, no harm done.

It would nice to see something along the lines of BBC Backstage, where BBC data is made available and individuals are encouraged to use it in innovative ways and share the results.

It’s all speculation at the moment, but the fact the the OS is sponsoring Geograph British Isles has got to be a good sign things are changing.

I wonder if the “2.0″ moniker is likely to be worn a bit thin this year? What it means depends on who you ask. If you ask me, I’ll start channelling Douglas Adams and refer to the fundamental interconnectedness of all things. Whatever. It’s great to see prominent technologists in organisations like the BBC and the OS waking up to it.

Pastebin performance improved

Pastebin never used to retain more than a few thousand posts, but the recent feature to allow longer post retention exposed a few performance issues. If you noticed pastebin.com being a little slow, my apologies.

I’ve made a number of fixes to address the problem, if you notice further issues, do let me know with a comment below.

Geograph, Creative Commons and Ordnance Survey – Oh My!

One of the reasons I started this blog was that I felt we were doing some interesting things with the Geograph site which wasn’t getting documented outside of private emails between the team. The recently announced sponsorship deal with Ordnance Survey is prompting a lot of activity in the team at the moment, so there’s lots to write about.

CC licence infectious?

Firstly, the press release attracted the attention of the OpenStreetMap mailing list, with list members speculating that the CC licence might “infect” OS maps used on the site. For example, if we overlaid images onto an OS map, creating a derivative work.

The answer is simply that we won’t be able to “mash up” an OS map with photograph contributions without the licensors granting us some additional rights. The resulting derivative, of course, could not be CC licenced. If it were, you could derive a clean, CC-licenced OS map from it. Wouldn’t *that* be something!

How about if we display an OS map on a photo viewing page to provide context?

No problem – the resulting page as a whole is not a derivative. The page is a collective work, and we’re free to include content on that page using different licences. The individual image remains CC licenced, but the entire page content is not.

We haven’t always got it right…

There are some problem areas we’ve been made aware of and tackled. We built maps using CC licenced images with coastline data derived originally from an OS source. As soon as it was brought to our attention we trashed the base map and rebuilt it from a public domain source (thankyou, US Government).

…and we don’t have all the answers

Can a CC licenced work include an OS grid reference? Oddly enough, this seems to be contentious issue. Is a grid reference a copyrightable work in its own right? I certainly hope not, and we’ve enquired with the OS to find out for sure. Some feel that if a copyrighted OS map is used to derive that grid reference, then the reference itself is copyrighted. To me, this feels like being told that if I refer you to page 42 of Hitchhikers Guide to the Galaxy, I’ve violated Douglas Adams’ copyright.

I can see it from the OpenStreetMap point of view – a series of grid references easily becomes a road, thereby reproducing an element of the copyrighted map. To use the book analogy, it’s like saying the first word on page 42 is “Ford”, the second word is “Prefect”, etc. These references are enough to reconstruct the copyrighted work.

A single grid reference applied to a photograph provides no reverse path for reconstructing OS property though.

Still, I’m no lawyer, and hopefully we’ll get an official line on that soon. If it proves to be a problem we can continue to the use the out-of-copyright New Popular Edition maps during photo submission.