A new failure in a CDN company has turned the internet upside down this week. Fastly is the company from which the fall of the pages of Amazon, Twitch, Reddit or CNN has emerged.
It is common for the internet connection to have a problem, something that is solved quickly and that leaves a page or social network down, but not so many or so important at the same time. That is why the alarm arose and the networks were filled with comments. BBC, Pinterest, The Financial Times or Reddit were some of those affected.
For a time there was speculation about an attack, there was even talk of a failure by servers like Amazon’s, but in reality the problem had arisen from a CDN (Content Delivery Network), Fastly company. These companies work to get the content of certain pages around the world more quickly and are a protection against some failures, but they are not infallible.
Actually, these kinds of simultaneous errors caused by a CDN are not new. Fastly’s direct competitor Cloudflare has already starred in more than one. In 2019 the dreaded 502 appeared on the screens of millions of computers and mobiles, Cloudflare had a problem and had taken many pages down.
While Cloudflare targets small and medium internet businesses, Fastly is focused on supporting large companies around the worldThat is why this ruling has been more popular. This CDN has explained that the failure was activated by mistake by a client of the company, triggering a massive drop in the other pages with which the company works. The failure had been hidden for a month and it was not until it was activated that the company’s computer scientists detected it.
The role of CDNs It’s of replicate your clients’ websites, its design, images and other elements for all the servers that it has distributed around the world. The Financial Times website was born in the United States and to reach the other end of the world with the lowest latency it is necessary to replicate it in a closer center.
In addition to reducing loading time for all users in the world, the chance of failure is decreased. If something does not work well on a server, users in a zone can access the second closest center and be affected, it is a safety net. But if the failure affects the structure of that network, everything falls like a house of cards.
Fastly lost 85% of its networkFortunately, they were able to recover the structure after a few hours. However, the time required and the number of pages affected reveals two things: that the failure was considerable, a significant hole in the structure, and that there is a very high concentration in this market for which many companies depend on a single supplier.
It did not matter that these companies will use servers from Amazon (AWS), Azure or Google Cloud, Fastly dragged them all. Even Amazon failed when it is a company that has its own CDN, as Jeff Bezos’ company prefers to decentralize so as not to further aggravate the consequences of a failure in its services.
Faced with a problem like the one experienced this week, users cannot do much, except wait for the company to find a solution. Be patient and check the social networks of these companies from time to time where they usually report the progress of the problem.