The Effects of the Fibre Outage Throughout the Mediterranean

Umar Kalim writes "Analysts have been studying the effects of the fibre outage throughout the Mediterranean in terms of network performance, by examining the changes in packet losses, latencies and throughput. We initially discussed the outage yesterday. 'It is interesting that some countries such as Pakistan were mainly unaffected, despite the impact on neighboring countries such as India. This contrasts dramatically to the situation in June - July 2005, when due to a fibre cut of SEAMEWE3 off Karachi, Pakistan lost all terrestrial Internet connectivity which resulted, in many cases, in a complete 12 day outage of services. This is a tribute to the increased redundancy of international fibre connectivity installed for Pakistan in the last few years.'"

Comments

Re:Who will benefit?

[palegray.net]

The question is really: Who would benefit from diminishing any country's Internet access during a time of war with that nation? Alternatively, conclusively proving that any nation's primary Internet backbone was destroyed might itself be the spark that ignites a war... who might benefit from that? Things get complicated pretty quickly.

Re:taps

[yidele]

This isn't an analog line, genius. The traffic of interest is in the client/ tributary payload, so the logical place to put a tap is where the container terminates or where it is switched. that means a telecom data center, not the bottom of the ocean. Additionally you needen't "splice" into live media ( thereby causing noticeable loss of traffic and noticably increased signal attenuation afterwards) - telecom transmission technology allows switching to protection at speeds 50 milisec, which is not noticable to human hearing...anyway, the data of interest would be on an encrypted data, not voice channel. 80GB is a lot of voice traffic, to sort this out in realtime you'd need a pretty stout set of telecomm eqpt. Consider that a single voice channel is 64kbps, also the fact that voice traffic ( if going over tdm and not voip) is multiplexed into sdh most likely carried over dwdm/wdm. A single e1 ( 2 mb/s) carries up to 32 voice channels, there are 70 e1 worth of payload in an stm1 , which is 155 mb/sec. How are you going to monitor this many discrete ( and multiplexed, encoded ) signals at once? do you even realize how much traffic that is? don't forget that sizable portion of voice goes over data ntwks as voip, and you can have ip using bridged ethernet over sdh, ima over multiple e1, mpls, atm or directly on top of sdh ( short list, there are many more possibilties)... anyway, if you want to listen in, you'd better do it somewhere you can actually do something with that data. having said all that, i think this is a good low tech way to test their readiness....