Fastly outage has also led to mixed outcomes

Traffic drop during Fastly CDN outage

Yesterday, Fastly made news headlines after a configuration error led to a wide-spread failure of commercial and critical apps and websites. This phenomenon has not only led to a new general interest in the inter-workings of the Internet, but also, as most news sources reported, exposed how fragile the Internet can be.

However, in a world as complex as the Internet, it is easy to miss out on the networks that have had a different experience than those underwhelmed by Fastly traffic during this event. As a network-monitoring-as-a-service provider (amongst many other things), we at BENOCS found that not all networks we monitor showed a drop in Fastly traffic. To show you what we mean, let’s compare two networks.

What we saw

In the image above, we can see that one of the networks showed exactly what we would have expected: a large drop in traffic (about half of the average value) with a deep dip shortly before 12:00 CEST, which matches Fastly’s reported time of 9:47 UTC (11:47 CEST).

However, when comparing that to a separate network in the image below, we actually see a sharp rise in traffic on that very same day with a deep dip around the time of the outage, followed by a major spike in traffic in the hours following.

Traffic spike during Fastly CDN outage

What could have caused the differences in these networks?

As much as we would like it to be, the answer is not so simple. One theory could be that an Internet event, such as an operating system update, a gaming event or download, or possibly a large streaming event could have occurred in the second network, which did not occur in the first. A second theory could be that traffic was off-loaded into the second network from neighboring networks. Whatever the cause, we can clearly see that the traffic in the second network is nearly double than its average traffic throughout the day, while the other network remains at roughly half of its average. Very unusual behavior.

What can we learn from these two networks?

The examples mentioned above show us that, when we look behind the curtain, the real-world is more complex and may often show us different or mixed outcomes than we originally thought. In a world as complex as the Internet, it is important to remember that not all networks will react to events in the same way, and it is important to be aware of what your network is capable of.

Are you looking for a smarter way to see your network traffic and its behavior? BENOCS Analytics may be the product for you. Get in touch with us today to learn more!

Back to media