NS1 gives our customers the unique ability to observe CDN outages and other performance issues as they unfold and to automatically steer around them, directing users’ application traffic to another healthy CDN provider. This ability to “see something, do something” in an automated way is paramount to keeping websites and applications online during such catastrophic events.
During the recent Fastly outage, we were able to help keep the internet running. This is what we and our customers were able to see and do:
The first graph shows the real user monitoring (RUM) data from our Pulsar® Active Traffic Steering product. Pulsar collects millions of RUM data points, which are latency and availability measurements from the browsers of real users across the globe to application endpoints. This provides a real-time view of the application experience as it would be perceived by the actual end user.
With this data, Pulsar generates a powerful real-time database from almost every geolocation and network ASN to those endpoints. This gives us the ability to observe application delivery availability and performance at a massive global scale.
As you can see, Pulsar observed a sudden drop in availability for Fastly. NS1 customers could see a shift in delivery between the hours of 09:59 UTC through around 11:00 UTC as Fastly started recovering.
Based on this data, NS1 customers who use Fastly as part of a multi-CDN approach had the ability to automatically steer application traffic away from Fastly at the start of the outage towards healthy, alternative CDN providers. Doing so enables organizations to continue to deliver performant application experiences to their customers.
This decision-making is done automatically through our powerful application traffic steering policy engine. Using our NS1 Filter Chain™ technology, NS1 customers can build complex application traffic steering policy rules that use real-time telemetry about internet, network, and infrastructure conditions to make optimal decisions.
The next graph shows the output of that automated decision-making process. As you can see, based on the data that Pulsar collects and analyzes, the traffic steering decisions were adjusted to minimize end-user impact from the outage. A significantly smaller share of traffic went to Fastly during the outage period, and then scaled back up as Fastly came back online. This required no manual intervention, which is especially nice considering this outage (as they often seem to) happened in the middle of the night for many of our customers.
As you can see from the graphs above, when a multi-CDN approach is implemented with intelligent and automated traffic steering, companies can mitigate the effects of even massive outages like this one. Pulsar gives organizations high observability of their global application delivery performance across multiple CDN providers, as well as the ability to immediately take action if a CDN doesn’t perform as expected.
NS1 Pulsar® Active Traffic Steering supports multi-CDN and multi-cloud strategies by leveraging real-time, global performance, availability and business metrics to automatically steer away from unwanted disruptions and latency issues. Having multiple providers in the mix allows you to dictate what may be most important: the most performant, meeting SLA targets, providing better coverage or simply steering away from provider outages.