The failure of Fastly, an American Internet cloud service provider, caused many official and private websites around the world to be paralyzed for nearly an hour, and this Internet corporate brand, which is quite unfamiliar to ordinary netizens, suddenly came into the public eye.
Fastly provides content delivery network (CDN) services to websites around the world to speed up user browsing speed and experience. During the accident, Internet users in wide areas of the world were unable to log in to many international media websites such as the BBC, the New York Times, and the Guardian, as well as online businesses such as Amazon, and even the websites of the British government.
Fastly announced the results of the accident investigation on the evening of Tuesday (June 8th) Pacific Time (West Coast Time) in the US, stating that the cause of the accident was that a program error (bug) was not discovered in a system update in May, and it was not a hacker attack. The company apologized for the paralysis.
Fastly is one of the few large CDN providers in the world. Information technology and network security experts have pointed out that this incident highlights the fragile side of the world‘s Internet infrastructure.
How did this network paralysis happen?
The incident occurred at 10:00 on Tuesday, GMT. When netizens tried to log in to multiple websites, the error message “Error 503 Service Unavailable” appeared.
Many sites are affected, including:
- British Broadcasting Corporation (BBC)
- British “Financial Times”
- The Guardian
- The New York Times
- Al Jazeera Media Network
Online forums, social platforms
- Shopify Canada
Fastly, headquartered in San Francisco, announced at 09:58 GMT that it was investigating an accident, and at 10:44 it announced that it had found the cause of the accident and was installing a patch package.
At 10:57 GMT, Fastly announced the completion of the amendment, but customers may have to deal with the influx of page views after the service resumes.
Fastly initially stated via Twitter that the company discovered a system setting that caused its global PoP (points of presence, network access point) to malfunction and has disabled the setting. PoP allows network content to be delivered to users from servers close to users.
On Tuesday evening, San Francisco time, Nick Rockwell, senior vice chairman of Fastly Engineering and Infrastructure, published a blog explaining that the entire incident originated from a software update on May 12, and a program error would be legally changed by the user. It occurs at a certain setting.
Tuesday’s accident caused 85% of the company’s network connections to report errors. Rockwell said that within 49 minutes, the repair team recovered 95% of the network operation.
“This is a widespread and severe network outage, which affects our customers and all those who rely on these customers. We are deeply sorry.”
The online post tag “#InternetOutage” (Internet Outage) quickly became a hot topic on the still functioning social media platforms. The increasing number of search-affected websites has surprised some netizens.
The media website can be described as the hardest hit area of this disconnection. Everyone tried every means to maintain news publication. Among them, the American science and technology news website The Verge used Google Docs to publish articles, but forgot to set editing restrictions. As a result, it evolved into a series of “chaotic entry” editors, which attracted onlookers on Twitter.
In an interview with BBC World News, Brian Barrett, executive editor of WIRED, praised Fastly for being able to repair such a major issue in just one hour. Paralysis accident. Some netizens seem to agree with this statement.
What is Fastly? What is CDN?
Fastly was established in 2011 to provide CDN services to major websites in the world to speed up the browsing speed and experience of netizens. There are not many large-scale competitors in the industry, including Akamai Technologies, Cloudera and Amazon Web Services (AWS).
Doug Madory, director of Internet analysis at the big data analysis company Kentik, pointed out to the BBC that CDN services are an important part of the Internet ecosystem. In fact, ordinary netizens deal with Fastly every day. “Usually you don’t know anything about it, because it is operating normally.”
CDN copies the content of the website to a network of servers all over the world, so that major websites can be downloaded to users more smoothly. It is especially important today when the website is full of high-resolution photos, videos and other content.
Dr Jessica Barker, co-president of the British cybersecurity company Cygenta, described on the BBC Radio 5Live program that the job of CDN is to reduce network friction.
Dr. Barker described the CDN like this: “In the real world, it is like a library. I can borrow a book from a library in the community without having to go to the publishing house to find it. But if it is a library When the door is closed, I won’t be able to borrow that book.”
CDN can help websites cope with the sudden influx of page views, and at the same time, they can help websites resist DoS attacks that attempt to paralyze websites by creating huge page views. Fastly uses a technology called “edge cloud” to do this, to the effect of dividing the data operations originally concentrated in the network center to local servers for simultaneous operations, reducing the overall network load.
How much damage did the accident cause?
We still don’t know how much financial loss this paralysis has caused to the world, but some people estimate that one hour of website failure can cost a company US$250,000; Reuters quoted multinational media public relations consultant Kantar as saying, according to A rough estimate is that for every hour of paralysis, global profitable websites will lose a total of $29 million in advertising revenue.
Professor Rebecca Parry of Nottingham Law School (Nottingham Trent University), a subsidiary of Nottingham Trent University, told the BBC: “The liability for service loss may be borne by the paid cloud service contract. But usually this kind of contract does not cover all losses.”
Fastly shares were listed in New York in 2019. It hit a high of $136.5 per share in October 2020, and has since continued to fall. After Tuesday’s disconnection, some netizens threatened to sell their Fastly securities.
Fastly shares fell 4% in the pre-market trading session on Tuesday, but did not fall but rose after the opening. As a result, it closed 7.7% higher than the previous day to 50.7 US dollars, which is equivalent to a market value of 5.87 billion US dollars.
Dr Martin Schmalz, associate professor of finance at Saïd Business School, University of Oxford, UK, described that netizens were surprised to find that such a huge network power was in the hands of a company called Fastly. The reader was pleasantly surprised to find that Fastly holds such a huge network power.
Can Fastly crashes be avoided?
Stephen Gilderdale, senior director of Dell Technologies in the United States, told the BBC that Fastly’s accident happened occasionally, but as long as the cloud service provider has enough buffers, the service will usually be normal soon.
Gilderdale said: “In most cases, the service will only be temporarily affected, and the data is easy to recover. It is far from worth worrying about. This can show the reliability of the network, and it can be restored in an instant if there is a problem.”
“Wired” magazine executive editor Brian Barrett pointed out to the BBC that other CDN providers have also encountered similar incidents. This incident reflects the current lack of adequate buffer security zones in the global Internet.
Corinne Cath-Speth, researcher at Oxford Internet Institute, UKPoint out on Twitter, Almost all websites in the world use CDN and other cloud services. This time there is a problem with the Fastly service, which has sounded the alarm of excessive centralization of Internet cloud services.
Reuters quoted the founder of cyber security company Idax Software and guest professor at the University of York, Prof Mark Rodbert, as saying: “It’s not easy that a single failure can cause chaos in the world in just 10 minutes. This makes it difficult. We realize that network technology to the cloud has changed the parts and links that enterprises need to protect.”
Professor Rodbert said that companies must build firewalls to ensure that the entire Internet will not be completely paralyzed by a company or even an employee.