India lost half its Internet capacity on Wednesday when two strands of fiber as thick as a thumb snapped. While service is returning to normal, it may be weeks before the cables are repaired. The fact that a pair of central Internet paths are just 2 km apart should serve as a cautionary tale as to just how fragile the Internet can be.
TeleGeography, a company that tracks cable capacity, said the two cables “account for the majority of international communications capacity between Europe and the Middle East.” With both cables cut, capacity has been reduced by 75 percent to only 620 Gbps. The cables also carry a considerable amount of traffic bound for India and Pakistan.
The Internet is designed to deal with outages. Its protocols are inherently resilient and can circumvent regional outages by finding new routes, but it can take hours for routers to agree on new paths around a breakage. Even once those new paths are established, they may not be as fast. In this case, the severed links were two of the fastest in the region, and now some data has to travel via satellite or across the Pacific rim.
We contacted Sef Kloninger, senior director of service performance at Akamai, to find out just how slow links became. Kloninger told us that for a 9-hour period — from 4 a.m. GMT to 1 p.m. GMT on Jan. 30 — tests from four cities in India across regular Internet connections climbed from their usual 2 to 6 seconds’ response time to as high as 15 to 30 seconds, and that those sites were only available to Indian users 50 percent of the time.
Akamai offers a service called “web application acceleration” that makes dynamic applications more responsive than they would be across such links. Kloninger said that even Akamai’s customers were momentarily disconnected following the break, but that unlike regular Internet users the company’s optimized routing algorithms detected the problem immediately and routed around it within minutes. “We saw most of the effect of routes into India, where connectivity was quite impacted,” he said.
India-to-U.S. response time is returning to normal. But Egypt’s minister of telecommunications, Dr. Tarek Kamel, expects the fix to take up to two weeks, and inclement weather is already hampering repair crews.
The fact that these two cables — which are supposed to provide redundancy — were so close together should be cause for alarm. When companies build data centers, they go to great lengths to ensure that power supply into the data center comes from multiple directions in order to avoid interruptions. But geographic and technical constraints mean some places are particularly crowded, much like modern versions of the Suez and Panama canals. And with this concentration comes vulnerability.
The cut also serves as a reminder of our increasing reliance on global internetworking. Customer service and offshore development are the obvious victims. But the Internet is so much a part of our lives that everything from Skype calls to family wire transfers were affected.
In May, hackers showed they could bring down an entire country. Massive botnets like Storm can control millions of computers with hundreds of thousands active daily. And a couple of simple cuts can leave millions digitally stranded and bring industries to a halt. As we come to depend on the Internet for more and more, we need to take a much closer look at its weaknesses.
Update: The BBC is reporting that a third cable in the Middle East, known as Falcon, has been cut as well. The Falcon cable won’t compound the problems of the previous two breaks, as the older Sea-M-We 3 cable is the one now carrying the bulk of data that Wednesday’s break interrupted. But conspiracy theorists are sure to start speculating.
Undersea Cable Map courtesy of Telegeography.