[Photo Galleries ][Polls Discussion ] [ Go to Castlebar ] [ The BB Index ] [ Disclaimer ] [Nostalgia Board ] [Roots ]
[NB Refresh for current version and all follow-ups]
Posted by Tech on March 10, 2004 at 01:29:11:
In Reply to: abuse of network by spammers, viruses, congestion problems posted by maybe on March 08, 2004 at 21:07:24:
To prove my point.
An article from the BBC News Website in the last few days. And it turns out that not just Irish subscribers are affected by poor investment in infrastructure and lack of redundancy. Even people in the beloved UK's lines go down sometime (NTL) making it not 100% reliable.
So M- Not to worry if castlebar.ie's mail is not available from time to time. It's not your fault.
Can the net take the strain?
I wanted to download the day's e-mail and check out my usual collection of news sites, weblogs and online magazines.
Although I have a phone that can collect e-mail, a lot of people send me large attachments which would take ages to arrive and cost a small fortune in GPRS data charges, so I do not read it on the train.
Even with Opera, the screen is too small and the connection too slow for any but essential browsing.
Sadly, I was deprived of my nightly fix of online content, because although the yellow light on my cable modem was blinking, my connection was down.
I could not even get an IP address for the laptop.
I have had my broadband connection from NTL for nearly three years now, and although it was unreliable for the first six months it has rarely been offline for more than an hour or so.
However, it was late and I decided I could live without my messages until the morning, so I turned off all the kit and went to bed.
It turns out that I had been fortunate not to be at home trying to do any work that evening, as the entire NTL network had gone down.
The problems were apparently caused by the failure of the TAT-14 undersea cable at around 1600 GMT on Monday.
The cable, which links the UK, several other European countries, and the United States, had already developed a fault at the US end earlier this month, and Monday's second problem knocked it out of commission completely.
France Telecom, who look after the European end, have sent a boat to dredge up the broken section and fix it, but it could take a while to find the right bit.
It is not clear to me why losing one link to the US should affect NTL's service so much, although it seems that the main problem was to do with the domain name system, so that customers could not find the websites or e-mail servers they were looking for.
But whatever the technical details, it meant that a lot of people were offline for around eight hours.
The cable break did not affect everyone. Although the cable is important, the way that the many networks which make up the internet are connected together means that data simply flows by other routes if one is unavailable.
These may be slightly slower, but even with a major link between Europe and the US out of action, few users should notice the difference.
I was lucky, because I have got a dialup account with Pipex which I use when I am away from home, and they were not affected.
I do not use NTL for web hosting or my e-mail, so that even if NTL's network had still been down on Wednesday morning, I could have picked up messages from my phone and updated my weblog using my dialup account.
But I am a sad geek who enjoys setting all this up. Most people rely on one company to do it all, and when that company fails to deliver it does not matter how resilient the rest of the internet is - they are offline.
There is a persistent urban myth that the internet was designed to survive a nuclear war and that it will be able to work even in disasters.
Even if this were true for the net as a whole, few of us have the multiple redundant connections needed to take advantage of it.
Renesys looked at how the August power blackout on the East Coast of the USA affected the internet.
They found that of the 9,700 internal networks in the area that lost power, over 3,000 suffered what they call "abnormal connectivity outages" - they went offline.
What is worse, many of them did not come back online even when the power did. Renesys found that more than 1,400 of the networks were affected for over 12 hours, and some for as long as two days.
Despite the fact that many companies and institutions, like banks and hospitals, rely on their communications networks and their computers, it is clear that not enough effort had gone into providing uninterruptible power supplies or battery backup.
Even though the high-speed backbone links were still there, the systems connected to them were out of action and company web servers or doctors' e-mails were unavailable.
This should worry us greatly.
The authors of the report believe they show that the net is not yet robust or reliable enough to take over from the phone network as the main communications infrastructure, and it is hard to argue with them.
The net is certainly a better-designed system than the power grid.
The cascade which caused the US power blackout to spread did not happen online, and internet connectivity outside the geographic area directly affected by loss of power was largely unaffected.
The high-speed backbone networks may be robust, but we should not blithely assume that everything will keep on working, especially when so many companies seem to be paying too little attention to how they stay online when the lights go out.