WeekEnd Feature: Bombproof?

by Ian Thompson, CCSP Staff Editor
April 3, 2004

If your information were vital, how far would you go to ensure it gets to where it is needed? How much would you invest to create a backup route? I know its all a matter of cost (or more specifically it’s a matter of cost-benefit analysis), but the costs are falling and perhaps it’s time to assess things again.

Mind that packet.

The Internet was designed to survive a ‘node failure’, which was a euphemism for ‘total destruction of a major US city’. As such, it can automatically route and reroute data to ensure that the complete message arrives at its destination regardless of the connections used. It does this by splitting the entire message in to smaller chunks (called packets), firing them out into the ether and then, by the power of friendly imps with a lot of Selotape, the messages are read and reassembled into their original order. Which goes part of the way to explaining why translation sites like BabelFish don’t quite make sense of a foreign language because the imps tend not to be cunning linguists.


Actually, I’m kidding: the messages aren’t fired out at random; they’re given a destination address, then handed to a team of small courier-imps that actually drive tiny taxis down the cable from your computer.


Anyone care to guess how hard my week’s been?


Where are we heading now?

Right – tangent time.


I’ve recently re-discovered Google’s News service. Considering it still proclaims itself to be a beta product, it does a remarkably good job of trawling the world’s news sites, compiling them into broad categories and then presenting them in a very clear way. I might even switch my homepage from my long-standing, fast-loading, never-wrong favourite, “about:blank”…


However, one feature it could do with is a ‘Watch this story’ option – where the reader could elect to keep the story for a few days, up to a week or so, in a ‘scrap-book’ fashion. This could be added to the Google Toolbar. Why? Well, it’s so good at finding stories that ones I spotted at the beginning of the week have now dropped off the page. And I remember spotting a couple of things that I might like to include here that I can’t find anymore. Oh well, maybe I should not be so lazy and just bookmark them!


Fire in the hole!

One of the stories I remember spotting here was about the recent fire in Manchester that took out a major fibre run provided by British Telecom that connected around 130,000 phone lines to the network.


This wasn’t just a case of households being affected – one of the major problems was that the emergency services in several regions were cut off from their networks. Apparently it didn’t affect the ability to receive calls, but (perhaps obviously) affected about 1/10th of the city, preventing them from making the calls. However, it also appeared to affect the inter-services communications and most ambulance crews in the area resorted to using their own mobile phones to keep in touch because their own radio system was dead.


Depending on your perspective, this problem was made worse because the fire also took out social services emergency lines provided to pensioners’ homes – they couldn’t call for help using this intercom-style system.


Clearly this also affected Internet systems; dial-up, broadband, leased services and the like. Call centres were shut down, company information clearly affected and life generally given a kick.


And this didn’t even take out the whole city – just one set of cables…


What have we learned so far?

Perhaps the most telling thing is that, clearly, the telecoms system in Manchester at least (and I’ve no doubt many other major cities, let alone more rural areas as well) has no capacity to deal with node or connection loss. This sort of thing must have been a commercial decision – to provide redundant, duplicated equipment has a financial impact. However, the likelihood of this kind of even occurring in a tunnel 30m/100ft below ground must have been judged so small that the cost-benefit analysis of installing additional capacity in another route, plus the equipment to control the data flow, weighed heavily in favour of not bothering.


However, is it really that difficult?


When the Internet was first created, back in the Sixties, the technology was clearly expensive, which is why it was essentially a military project. These days, the technology costs nowhere near the same.


Can’t see the wood for the trees.

Look up ‘STP spanning tree protocol’ on Google and you’ll find over 26,000 links. Some of them are very much linked to vendors like Cisco and Allied Telesyn, others merely list the protocol, whilst others actually explain what it is.


Here’s one: NetworkWorldFusion; and here’s another: Javvin (this includes links to a neat ‘Protocol Poster’ and also a packet sniffer – both to buy). My favourite literary source, O’Reilly, offers this - but alas without the usual animal sketch cover…


The thing is, STP offers us mere mortals the kind of resilience to failure that the Internet has enjoyed for over 30 years. Many ‘professional level’ items of networking equipment have this feature, including the Allied Telesyn Rapier G6F switch fitted in the centre of my school network. It’s a feature that takes some setting up – things like path priority and so on need careful tweaking to prevent a packet storm form building up. This is were the imps all drive around in circles, ever faster, with ever more tiny taxis, like some mini-M25 orbital motorway, packed tighter and tighter until BOOM! One of the imp drivers loses it and the whole circuit grinds to a halt.


Actually, it’s where the redundant route is seen as easier to ‘travel’ down for the data packet than the off-ramp to the destination PC. Eventually, there’s so much traffic just circulating around the main and redundant pathways that there’s no capacity to accept anything else and the network locks out servers, end users, etc.


So, a Leeds school is better equipped than central Manchester…


I guess you’ll find that many places are. Most companies that rely on networking will have created fairly sophisticated LAN structures, including my school. Personally, I have experienced a series of failures in the old network that I was keen not to ever see again. When it came to specifying the replacement last year, I not only asked for redundancy in the cabling, but also in the main servers. It was an unusual move last year for a school to specify a wholly redundant server (in terms of the services it offered end-users), but it isn’t so odd for companies to do this. In my case, the extra £4k was to ensure that I didn’t suffer a total failure in both PSUs of the main PDC server (as happened on the old system) and lose the network for a month whilst spares arrive.


I note with interest that other schools are now following suit, or at least considering it. Most will be as a result of a failure in a hard drive or mobo, or some other vital section of the core systems.

Similarly, I didn’t want some contractor come on site and put a spade through a cable – the network had to survive this and STP was employed. Actually, the cunning plan of putting the cabling in the ceiling should reduce the likely attack by wayward digging implements, but you get the idea.

Fibre cable costs very little to install. If STP is included in network switches, then why not make use of it? At least it would help cope with any on-site, internal catastrophe.


Perhaps this sort of question is going through BT’s corporate mind as we speak? Along with “Just how much do we owe to customers for this loss of service?”, of course…



by Ian Thompson ComputerCops Staff Editor



Ian Thompson is a Network Manager of a 500-PC, 9-server, 1700-user school network and is an ICT teacher at a UK high school near the city of Leeds. He has written articles for the Hutchinson Encyclopedia, plus many resources in support of teaching ICT in the UK schools' National Curriculum.


Copyright © Ian Thompson All Rights Reserved 2004.