Bird Flu Pandemic Could Choke the Net

PetManimal writes "If a pandemic were to occur, many companies and organizations would ask their staffs to work from home. The impact of millions of additional people using the Internet from home might require individuals and companies to voluntarily restrain themselves from surfing to high-bandwidth sites, such as YouTube. If people didn't comply, the government might step in and limit Net usage. The scenario is not far-fetched: last year at the World Economic Forum in Switzerland, a group of telecom and government officials conducted a pandemic exercise based on a hypothetical breakout of bird flu in central Europe. The results weren't pretty." From the latter article: "'We assumed total absentees of 30% to 60% trying to work from home, which would have overwhelmed the Internet,' said [one] participant. 'We did not assume that the backbone would be gone, but that the edge of the network... would be overwhelmed... The conclusion [of imminent collapse] was not absolute, and the situation was not digitally simulated, but the idea of everyone working from home appears untenable,' [he] said."

Re:Why

[Anonymous Coward]

Maybe because of likely recommended or enforced government quarantines or other advice for people to avoid unnecessary contact with each other in an attempt to try to stop or slow the spread of the disease, which will be airborne and spread at places where people congregate?

Just a guess.

Re:Why

[timmarhy]

because most epidemics are spread via work places and public area's.

Re:Why

[TempeTerra]

I'm not sure what you mean by that. Working from home is perfectly sensible in case of an epidemic, although I'd be inclined to ditch work altogether ;) . One of the first things to do is close all the schools so kids don't share their germs around. Non-essential businesses are the next to go.

Re:And a butterfly could cause a hurricane

[Karganeth]

Haven't you heard of the the butterfly effect [wikipedia.org]?

Re:9/11 caused net stoppage

[foobsr]

people dying in their beds

Worse, if like the Spanish flu they will probably be dying in the street - as a friend who has learned it from his father who was an eye-witness told me - and is also mentioned here [historysociety.ca], quote: "Victims were dying in the street, in stores, in offices, in military barracks, turning blue and struggling for air as they suffocated in bloody froth.".

Reason enough for people to use youtube just for the sensation.

CC.

Re:9/11 caused net stoppage

[Bastard of Subhumani]

I reasoned that the BBC's website probably wouldn't be heavily flooded with North American traffic, and that it would be the middle of the night on that side of the pond.

The 9/11 attacks happened in the morning (local time), which is early afternoon in the UK.

Re:9/11 caused net stoppage

[locofungus]

WTC stuff: because I reasoned that the BBC's website probably wouldn't be heavily flooded with North American traffic, and that it would be the middle of the night on that side of the pond.

We're five hours ahead of you, not behind you. It was early afternoon here when the first plane hit.

Tim.

Re:Why

[Bloke down the pub]

... yet. The reason there's been 'only' 275 cases so far is that humans are catching it direct from birds. If (some say it's when) it mutates to the extent that human to human transmission occurs, it's an entirely different matter.

Re:9/11 caused net stoppage

[Anonymous Coward]

I reasoned that the BBC's website probably wouldn't be heavily flooded with North American traffic, and that it would be the middle of the night on that side of the pond.

Eh? Did England move West when this happened? I was watching a CNN feed of this (we had no TV available) when this happened at lunchtime whilst at work in the UK!

Also, the BBC site was useless then but we could get a CNN feed? Strange...

Re:And a butterfly could cause a hurricane

[Shakrai]

That the net is inherently able to route around problems is obviously ignored here.

If that problem is a flood of unanticipated traffic then where it is it going to route to? And most routing works on a shortest path first basis. If that path is congested then the packets start to go into queues. They don't magically take another route (in most routing configurations).

Anybody remember 9/11? I can't be the only one that found many services to be borderline useless that day. Our backbone wasn't even maxed out and I still issues using VPNs between our offices (which weren't maxed out either). IM, various websites (the news ones), IRC. They were all sluggish and non-responsive at times.

Re:And a butterfly could cause a hurricane

[msobkow]

I think you're underestimating the potential risk. A pandemic is far more likely than a major terrorist attack or any other such nonsense causing hundreds of thousands, if not millions, of people to work from home. Businesses could not just shut down were there a pandemic worse than SARS.

When SARS hit the GTA, there was a significant increase in remote access to corporate resources from telecommuters. But while this article focuses on the impact on the backbones of the internet and the potential need for data- and site-based traffic shaping, it neglects to consider the far greater risk of individual businesses which flat out do not have the connection capacity to have the majority of their employees working from home.

Just because risks are low doesn't mean problems cannot happen, and a good business manager needs to allow for those risks. Consider something so simple as a RAID-5 disk array. Most techies consider them virtually fault-tolerant and bullet-proof, yet I personally know an admin who had a second drive fail while replacing a bad drive, losing the whole array.

That site now uses RAID-6 (two parity stripes instead of one) so that they reduce the chances of losing any of their servers in such a fashion again. Yet even they know it's only a statistical game and that it is theoretically possible to have three drives fail at the same time. There are just limits as to how much you invest in hardware to avoid such problems before one starts looking at full off-site redundancy solutions that cost millions, not thousands of dollars.

If you want a US-based real world example, take a look at what happened to industry on 9/11 and the subsequent week. I worked for a company that lost people, hardware, and services that had been operating out of the towers. The impact was not small, and if we hadn't had disaster recovery plans in place and tested ahead of time, the impact would have been much worse.

You're free to stick your head in the sand and ignore risks, but some industries (such as banking) don't have that option.

Re:And a butterfly could cause a hurricane

[petecarlson]

Although, to a point, I am also a little irritated at how shared bandwidth is marketed, you know damned well, without having to read the fine print, that you are buying shared bandwidth, if you are paying less then $100 per Mb/s per month. All this crap about ISPs selling best effort bandwidth drives me batty. If we all refused to sell shared bandwidth and made you pay for that 5/1, You would be paying $600 - $1000 per month for the connection and bitching up a storm about it. If we sold it to ten people and they each payed $60 - $100 per month, you would be bitching that it was oversold. One way or another, someone needs to pay for the connection. The market has come up with multiple ways to buy bandwidth. Chose what you are willing to pay for.

Back to the topic.
Disaster, Bird Flu or whatever, the first thing to go on my network is best effort bandwidth. If needed, I will throttle it back to ISDN speeds before I even think about touching an SLA account.

Re:And a butterfly could cause a hurricane

[twiddlingbits]

The major routers on the Internet are setup to provide many alternative paths based on congestion or other sorts of delays. Yes, they always try shortest path first. But, they don't just try one route and say "I give up, lets queue these packets". Some in fact have very clever algorithms to meet QoS standards via many different alternative routes. Some corporate networks do as well. You also can assume that while most of Europe is relaxing at night (lower-bandwidth) most of North & South America is working, and when the Americas are off-work Asia-Pac is in prime work hours. So there will only be a few times when everyone who is a heavy hitter is online together. Also high bandwidth sites can implement throttling where they don't feed as many users or they feed less packets to users to help bandwidth usage. I'd worry a lot more about the external interfaces to corporate networks choking before I would worry about the entire Internet. Plus the telcos have massive amounts of dark fiber they can turn on within a very few days (left over from the dot bomb build it and they will come days). Worse case the congestion lasts a few weeks, but it won't bring the world to a halt. This article is not well thought out, in fact it may have even been funded by bandwidth providers. Mod post down.