Update (15:26 UTC, 15 May): Routes to Syrian networks have been restored, at 18:26 Damascus time. Outage duration: 8h25m
Update (14:20 UTC, 15 May): Plot of latency measurements to Syrian hosts from various locations, indicating that replies stopped returning shortly after 7am UTC, aligned with the withdrawal of routes to Syrian networks. (Click image for details)
Update (07:30 UTC, 15 May): Syrian Internet down again since 07:01 UTC (10:00 Damascus time), Wednesday, 15 May 2013. Syrian news agency reports that they’re working to fix. Potentially related to forthcoming UN decision today?
Update: Syrian Internet has returned. Outage lasted 19.5 hours, from 18:45 UTC May 7th to 14:13 UTC May 8th.
As we write, the Syrian people are still disconnected from the global Internet at the most fundamental level, nearly all of their paths withdrawn from the global routing table. Since 18:45 UTC on May 7th, Renesys hasn’t seen a flicker of activity. We haven’t been able to successfully send a ping or a traceroute to any host inside Syria. Government websites, universities, domain name servers, core infrastructure routers, banks, businesses, DSL customers, smartphones: all silent.
As I look back at what we’ve written about Internet outage over the years, I see a sort of evolution in our perspective. We’ve covered Internet failures due to war, politics, censorship, central planning, earthquakes, hurricanes, cable cuts, business disputes, terrorism, undersea mud volcanoes, and (perhaps) cyberwarfare.
In the early days, we reported each outage breathlessly, shocked that the Internet could fail in such spectacular ways. If you look around the web this morning, you’ll see a lot of that same shock-and-awe reporting from companies who are just discovering the fragilities visible in Internet data.
In this case, however, what strikes me is the depressing sameness of the sequence of Syrian Internet disconnections. Just as in June 2011, July 2012, August 2012, and November 2012, the entire nation disappeared from the Internet in 30 seconds, as if a switch had been thrown. Everyone in the Twittersphere seems to share the same strange lack of perspective about these events — in the middle of the chaos and tragedy of civil war, why is anyone surprised when the Internet stops working?
Isn’t it actually more shocking and noteworthy that the Internet in Syria actually functions pretty well 360 days out of the year?
If you’re looking for a different way to think about events like these, you might enjoy reading Nassim Nicholas Taleb’s book, Antifragile (ironically, published just days before the last major Syrian Internet disconnection).
Antifragility is a property of systems that get stronger when they are challenged by stresses. It’s not merely robustness, or resilience (systems that return to stable state after stresses have passed) — an antifragile system like the Internet actually gets better when you try to break it. Systems of systems attain antifragility because their individual elements are so fragile. For example, Taleb writes:
“Restaurants are fragile; they compete with each other, but the collective of local restaurants is antifragile for that very reason. Had restaurants been individually robust, hence immortal, the overall business would be either stagnant or weak, and would deliver nothing better than cafeteria food — and I mean Soviet-style cafeteria food. Further, it would be marred with systemic shortages, with, once in a while, a complete crisis and government bailout. All that quality, stability, and reliability are owed to the fragility of the restaurant itself.”
If you’ve been reading all of our Internet outage case study blogs, you might see a familiar pattern. The Internet fails in some way, taking individual institutions (or even entire countries or geographic regions) offline. The outage exposes fundamental undetected weaknesses — lack of submarine or terrestrial cable diversity, lack of Internet provider diversity, underinvestment in alternative forms of transport, political control over Internet chokepoints. If the Internet damage is painful enough, engineers (or revolutionaries) are forced to take creative steps to address those weaknesses. Repeat… forever.
Take Pakistan, for example. This cycle of insult-and-repair is what made Pakistan Telecom serious about buying eastbound backup transit to Asia in the wake of the first round of Mediterranean cable cuts in 2008. It’s what led (indirectly or directly) to the establishment of competing providers in Pakistan, with foreign investment and additional cable resources connecting Pakistan to the Gulf States.
And because Internet is easily exportable, Pakistan’s improved Internet connectivity spilled over to improve the situation in their historically fragile neighbor, Afghanistan. Here’s a picture of one slice of Afghanistan’s diverse connectivity over the last few weeks (click to zoom):
Note three paths from Afghan providers to Cable and Wireless in Europe. Borrowing from Pakistan’s diversity, one route goes through Pakistan’s competitive provider, Transworld Associates, then to Omantel and over the EPEG cable through Iran and the Russian Caucasus to Europe. Another goes by satellite to Bahrain’s Batelco, then over to Omantel and out through Iran on the same EPEG route. A third goes north through Uzbekistan’s Intal Telecom, then through two Russian routes to Europe. The Iranian EPEG route is itself an improbably antifragile innovation, inspired by multiple years of Mediterranean cable cuts and avoidance of the demonstrably fragile Egyptian transit corridor.
So what does all this mean for Syria and its Internet neighborhood? One has to look past short-term dysfunction and think about what comes next. Every significant Internet disconnection, and the local and global reaction of outrage and dismay, sends an important signal about the fragility of the underlying system. It makes single points of failure and control visible, so that those fragilities can be found and fixed, and the Internet as a whole can continue to gain strength from disorder.