Early last month, my blog "Pinning Down Latency" included this prediction:

In the coming weeks we expect to see a dramatic shift in transit as Lebanese providers move away from expensive and high-latency satellite service to IMEWE-based service.

Well, it didn't take long for this to play out.

We can confirm reports of significant but sporadic Internet outages in the Palestinian Territories today.  As many as half of the routed networks of the Palestinian Territories were unreachable (withdrawn from the global routing table), possibly as a result of reported cyber attacks.  These outages are the largest we have observed all year for this country, which normally has a fairly stable Internet.  Impacted networks are located in both the West Bank and Gaza Strip.

 

For about 3.5 hours today, almost half of the Internet in Pakistan was down. Renesys observed 46% of the routed networks of Pakistan withdrawn from the global routing table between 19:37 and 23:02 UTC making this the largest Internet outage we have observed in the country in recent years.

Undersea cables are expensive to install. But if you're an Asian Internet hub trying to connect to other Asian Internet hubs across un-cabled waters, what else can you do?

Well, one alternative we see is Internet Providers heading to California, as many Asian providers opt for Internet paths out of Asia to the west coast of the US, and then back to Asia. These tortuous routes, aptly called hair-pinning (observe their supple shapes), may be cost-effective initially, but generate latency, which can be a problem for some businesses (and their end users).

Packet latency is a big issue in Internet-based applications (i.e. the stuff in the cloud). In conducting analysis on Internet infrastructure over the years, we have seen many patterns of connectivity. One such pattern that can wreak havoc on latency is "hair-pinning", a phenomenon where traffic takes an unnecessarily long physical path between two points on the Internet due to suboptimal routing. The increased distance results in increased latency, and the "lag" or "sluggishness" that users experience as a result can hinder latency-sensitive online applications whether they are financial trading applications or MS SharePoint.

Hurricane Irene knocked out power to millions of homes and businesses as it travelled up the US East Coast this weekend. Even as the winds subsided, torrential rains triggered savage flooding throughout Eastern New York state and Vermont, tearing up roads and exposing the telecommunications infrastructure to further risks. The storm's impacts were clearly visible in the Internet's global routing table, as tens of thousands of networks were cut off from the rest of the world.

Here are a couple screenshots from our Internet Health Portal, which we provide to the US Computer Emergency Readiness Team (US-CERT). During an emergency like Hurricane Irene, this tool provides the US-CERT with critical information about the availability of Internet services across America. Working from lists of impacted customers in each state and county, and lists of correlated outage events, we can supply a lot of useful information about the problems being experienced by enterprises in the affected area. That information can be passed along to state and local governments to aid in prioritization of disaster relief.

As dawn broke in Libya on the morning of Sunday 21 August, it appeared that the battle for control of Tripoli was underway. Throughout the night, a steady stream of tweets and retweets emerged from Libyan sources, painting a confusing, often contradictory picture of the evolving situation.

We're still piecing together the data that can confirm or deny much of what's been reported overnight, but one thing is clear: something very strange was going on with Tripoli residents' Internet access. Service was restored suddenly in Tripoli, flickered on and off for a couple of hours, and then died, with the majority of the country's international BGP routes withdrawn from service for good measure. Today the routes are back in Tripoli, but ADSL service isn't. This morning we're looking back at this curious overnight episode, and speculate about what might have happened.

There hasn't been much to say about the Internet in Libya this summer, as their patterns of connectivity have been fairly stable. It was interesting, therefore, to observe that much of the country's Internet routing has started to show evidence of sporadic failures this week, which have gone unreported in the media.

The following plot shows the number of Libyan networks (blocks of Libyan IP addresses) that appear in the global routing table. There are typically 16 of these, all routed by Libyan Telecom and Technology (LTT) via Telecom Italia. This week they have suffered some impairment, in groups of 6 or 10, in episodes that typically last no more than a few hours.

There didn't seem to be any pattern to these outages, which took place at all times of the day and night. It seems to suggest power outages, rather than permanent facilities damage, or deliberate action by the government (for suppressing communications, say).

Is it possible that LTT is suffering power outages, and having trouble finding fuel for their generators due to NATO's unofficial fuel blockade of Tripoli?

One reason why the world at large may not have noticed much in the way of Internet impairment: the affected networks don't seem to be the same ones used to access the majority of Internet content from inside Libya. Looking at Google's plots of inbound traffic from Libya, these substantial network outages seem to have had very little impact on the daily traffic curves:

In other words, the handful of Libyan networks that aren't affected by these outages seem to be the same ones that are consistently generating the Google traffic. If you're lucky enough to be an LTT customer in one of these Internet neighborhoods (presumably in Tripoli), your connectivity stays up. Everyone else, well, you're on your own. Time for Internet in a suitcase?.