This article is more than 1 year old

Reg Oz chaps plot deep desert comms upgrade

We'll be heading back to the remote community of Willowra in 2015

It's been a while since The Reg travelled to the remote Australian community of Willowra to perform a Windows XP upgrade, and longer still since we updated readers on our plan to improve the internet connection in the town's adult learning centre.

We'd like to fix the internet problem because the Wirliyatjarrayi Learning Centre is making a difference: little things like online banking make a difference when the nearest bank is five hours by (tricky) road.

But of course things aren't as simple as we'd like.

We'd given some thought to the kinds of tools that would help us collect information about the kind of traffic traversing the network, without doing anything that might compromise the privacy of end users (there's no point in trumpeting peoples' right to privacy and then ignore it ourselves).

The most popular suggestions were to analyse DNS requests, or to run a tool like Wireshark or ntop to get a feel for the big traffic types, but a discussion in Sydney with Riverbed engineer Nathan Chan offered the possibility that perhaps we could deal with network optimisation without having to try our own mug-shot at network traffic analysis.

So we sat down with Chan for an hour with coffee and a whiteboard and sketched out what we knew about the network – and that's where we ran into a possible roadblock: the WiFi design.

Since the user management – including clicking “I agree” to the content filter all users must log into network terms and conditions when logging into the network – is handled by a third party, it's feasible that the logins happen not in Willowra, but in a network controller hosted by a contractor in Melbourne.

That makes things a lot more problematic, in terms of network optimisation, because it means we have lots of extra hops to deal with to improve the connection.

Here's how we think the current path looks:

User – WiFi access point – Satellite modem – Satellite – Ground station – content filter at contractor – Internet.

We'd prefer a path that looks like this:

User – WiFi router – Satellite modem – Satellite – ground station – Internet

If we have a neat path between the user and the Internet, then vendors like Riverbed can help without too much difficulty, because a WAN optimisation box in Willowra would have access to the traffic in its raw state.

If, on the other hand, the access point is just a piece of dumb kit with the smarts somewhere else, the traffic between will be encrypted and not accessible to optimisation. A proxy between the network and the access point will give users some relief, as would doing a better job of patch management.

So El Reg is continuing to seek data, in particular a better understanding of the specifics of this implementation, and we're keeping an eye on key calendar dates for 2015 to see whether there's a chance to get more local control over the implementation, without sacrificing the filtering and user management that Batchelor Institute needed outside help with.

The good news is we've been able to fire up tracert in Willowra, so now have a network path we can analyse in detail to figure out just what lands where, where the horrid latency comes from, and seek information to help us figure out what to do next.

We've also secured a couple of very small PCs on which to build optimisation appliances. Watch this space, readers, there's plenty more news of this project to come in 2015. ®

More about

TIP US OFF

Send us news


Other stories you might like