Ahhh...
That explains it... I noticed some weirdness yesterday, but given it was on a friend's Talk-Talk connection I didn't realise it was anything unusual!
So where's the Sun headline...?
Undersea internet cable severed, Europe cut off.
A major internet routing outage struck UK telcos over the bank holiday weekend - knackering access to the World of Warcraft website*, the BBC, Amazon, Facebook and other sites for more than 24 hours. It's understood that a submarine cable carrying web traffic snapped between Blighty and the Netherlands, causing headaches for …
It requires LOS, cannot work over a curved surface (which the earth is) and also no obstructions or interferers.
Throughput(Guided (Fibre) Optics) > Throughput(Unguided Optics)
at any moment in time.
Doing optics without the fibre and calling it wireless isn't really a substitute, which is what I think you are trying to say?
They're the only thing available to mobile stations like ships or "internet to a field/beach in the middle of nowhere". (Or oddly Hyde Park a few years ago, before 3G took off)
Approximately 500ms ping is normal. In some cases I've had to adjust timeouts to get a connection to stay up - under poor signal quality roundtrip time goes up quite a bit due to retries.
I rather doubt you'd be happy over that one.
"Hello, I don't have a clue how the Internet works or any understanding of transmission technologies, bandwidth or spectrum. Strangely though, I feel empowered and emboldened to offer my suggestions for improvement in front of a group of people who are probably networking experts."
Does that about sum it up?
Okay, tell me that it's not just me.
The point of running huge data centres with failover hardware, and the point of IP routing is basically so that, should something happen, the dead bits die and everyone is automatically shifted onto the non-dead bits.
Assuming that the capacity of your average ISP isn't being constantly tested 24/7 on all of its lines (which kinda makes the word "redundancy" not applicable), why should any ISP have to take any physical action for this kind of thing to be noticed, monitored, and then failed-over to other routes? I can understand warnings, and notifications, of it happening but why does it require manual intervention? Why are there half-a-dozen routing protocols professing automatic best-route selection if nobody uses them? Why can I set up a simple fail-over over several routes in about two lines of "ip" commands and yet big ISP's, datacenters and network providers can't do the equivalent on their expensive Cisco equipment?
I completely understand "capacity problems because we're running on one-line-less than normal" but "takes 30 hours for someone to work out what to do and find another hole to plug the cable into?".
Do we really have an Internet where manual intervention is required to provide the routing that's required when something fails?
Hmm - capacity on this scale isn't "Pay As You Go". It's contracted for 12 months or longer and failover protection will be bought or not based on a simple sum based on the cost of the protection versus the cost of lost business if the primary route fails for the maximum time allowed in the SLA.
Re: Do we really have an Internet where manual intervention is required to provide the routing that's required when something fails?
The outage affected/affects non-protected pathways - i.e. there was no redundancy, probably to save on costs.
I'm guessing that all of the affected parties that required manual re-routing did so because using alternative routes costs them more.
>Do we really have an Internet where manual intervention is required to provide the routing that's required when >something fails?
No - you can have an internet where the router has the company credit card and will go and buy capacity on whatever link it needs, automatically at whatever the cost.
It can happen. If someone else's kit down the chain on your alternate route thinks the best path to your destination is back the way you just came, those packets evaporate. As an individual ISP you don't own all the routers on the way to where you're going, other people's routing decisions determine how effective your back-up is.
You could build your own private backup, totally under your control, but £15 a month broadband doesn't pay for that.
Likewise. I've been procrastinating about moving from BE to 40MB SKY as BE have been so good in the past. But the have lost their way so regrettably it's time to help line Murdochs pockets. The only saving grace I had was I have a VPN option on my Usenet account. I've never used it before but it worked a treat and my WoW latency was at least 10ms lower than BE's standard.
Sigh. And if the other links are running at 80+% of capacity? Traffic will be rerouted, but then the other links are saturated.
Idiots analagy. Assume you work in London and drive to Hemel Hempsted or Milton Keynes, and both the M1 and M11 are closed. Yes, you can divert. But it's not going to be fun…
In this case my ISP's connectivity was half-fucked, the routes to the VPN(s) and then on to certain destinations were absolutely fine, yet the direct route to the certain destinations was not always fine.
The only downside of my VPN access is that it bypasses the very helpful ADSL firewall so I have to beef up network security on the PC to stop the casusal 'hackers' (loosely put, probably bots) from trying to connect to certain ports random IPs.
Took ages to convince first line support there was a problem, which involved switching my router on and off again ad infinitum.
Could tell something was broken by looking at traceroutes and comparing that with my not-broken 3G connection.
Finally got through to second line support, took my details, and told me they'd take a look at it after the bank holiday. Nice work guys.
They just ripped up some cables when they were laying some new ones.
See recursion.
http://www.outpost9.com/reference/jargon/jargon_33.html
Submarine Commander:
Yes, sorry 'bout that. Slight tits up. Ripped up some cables when laying some new ones. Shame really.
Should be ok by morning. Yes, yes, yes, we shall relay all the old cables being ripped up. Unless we rip up any of the new ones... when we are relaying the old ones. If you follow me, sir...Or fail to lay any of the new ones whilst replacing the old ones. Could get a bit sticky sir.
Sorry!
Anyway, send us a line, if you can get through sir. Yes, yes, old cable's as good as new cable....
(mutters under breath) if you can find any fucking cable left by the time we have finished (/muttering).
At least they didn't have the problem the HMS Banana had. I heard they went six times around the world before they finally run out of fuel.
See: banana problem /n./
See:http://www.outpost9.com/reference/jargon/jargon_17.html
Talk about redundancy.
The lose of a cross channel is common event. It is common enough to have several specialist ships based on the North coast of France just for doing the repairs. Common route protection schemes will detect and reroute sub-second what ever the tech. Even the slowest of routing protocols can figure it out in a minute or two.
Interroute fibre is meshed. So it does not re-route you it is a commercial not technical issues. It may not be expressed that way in the contract, but premium clients will get resilience. Less valueable clients get less resilience. You get what you pay for.
"So it does not re-route you it is a commercial not technical issues. It may not be expressed that way in the contract, but premium clients will get resilience. Less valueable clients get less resilience. You get what you pay for."
Usually it is expressed in no uncertain terms in contracts if a customer is buying protected or unprotected services. Customers may not always understand the implications, or even the technology options if they're unfamiliar with the way wavelength services are delivered, ie wavelengths are generally dumb clocked light. That can be especially critical when buying submarine capacity when fixes often take more than a truck roll. Another common problem is trying to sweat the asset. Two links running <50% good, >50%, not so good in the event of losing one.
Hibernia Atlantic made some videos showing aspects of submarine cable laying and repairing. Basic fibre splicing is the same as for dry/terrestrial cables. Tricky bits come in dealing with armoured cable and the power. This video shows how jointing is done-
http://www.youtube.com/watch?v=dIa-kyYndwQ&list=UUoR_9pS8jCcf-eEArrJnI6Q&index=53&feature=plcp