Internet Exchanges (IX) play an important “efficiency” role in the practical matter of physically connecting the networks that comprise the Internet. IXs are facilities that provide space, power and connectivity to member networks—typically Internet Service Providers (ISPs), telecom, cable and mobile carriers, content providers, web enterprises and sophisticated enterprises such as large financial services companies. This infrastructure is used by these networks to arrange peerings exchange regional, national and/or international IP traffic. By being able to peer with many different networks in one facility, networks can cost-effectively gain lower-latency connections than if they had to reach those networks via multiple transit hops. The availability of multiple peerings also increases network and transit path resiliency. There are hundreds of for-profit and non-profit Internet Exchange Points (IXPs) in the world, including giant operators like Equinix as well as organizations such as AMS-IX, LINX, Netnod and DE-CIX. For reference, here’s a nice directory of internet exchanges hosted by Packet Clearing House.
So when a major IXP has an outage, the effects can ripple pretty far and wide. This is what happened on April 9th, 2018, when a power outage took down Interxion’s FRA5 datacenter and the DE-CIX switch that many networks use to connect to DE-CIX’s distributed Frankfurt IXP. In addition, the Interxion FRA5 switch outage also affected DE-CIX’s own operational systems including their email servers, leaving them only Twitter to communicate with their customers, as seen in Figure 1.
The initial power outage lasted for about an hour and a half in the mid-afternoon. Power was restored, then lost again in the evening for another couple of hours.
The Ripple Effects of the DE-CIX Outage
The impact was felt acutely, especially in Germany. In general, DE-CIX Frankfurt is a large exchange—with around 800 different networks peering through a distributed fabric and pushing around 6 Tbps of traffic. According to media reports, German Internet connectivity was broadly affected.
At ThousandEyes, we were able to capture and visualize the impact of the outage on a variety of large Internet transit networks, as shown in Figure 2 and Figure 3. The red-outlined circle show provider networks that were directly impacted, with a number signifying the number of impacted interfaces—which most likely correspond to how many connections each provider had to the fabric at DE-CIX6/FRA5. The ripple effect beyond those directly affected ASNs is shown by the fan out of other ASNs that are connected to the directly impacted providers.
Why Understanding Internet Outages Matters
Nearly every modern business already has or is building critical dependence on the Internet, whether it be via engaging customers and partners with external-facing websites and mobile apps, building micro-services in IaaS clouds, or utilizing SaaS. The growing dependence on so many Internet-connected resources means that understanding Internet changes, outages, instabilities and trends need to be part of the core intelligence for IT departments. If you’re running blindly to Internet dynamics in the age of the cloud, it’s high time to get more tuned in and turned on using Network Intelligence. With digital experience being so existentially important, how you play the Internet game may well be a make or break matter for your business. Feel free to request a demo if you’d like to learn more about Network Intelligence.
