[ad_1]
That’s not all! CDNs don’t just store content closer to the devices that need it. They also help guide it through the Internet. “It’s like coordinating traffic flow on a huge road system,” said Ramesh Sitaraman, a computer scientist at the University of Massachusetts Amherst, who helped create the first major CDN as Akamai’s chief architect. “If some links on the Internet are down or congested, the CDN algorithm will quickly find an alternative route to the destination.”
Therefore, you can begin to understand how a CDN can take up a lot of the Internet when it fails. Although it does not fully explain how far-reaching Tuesday’s impact was, especially when so much redundancy is built into these systems. Or at least, there should be.
CDN integration
Again, it is not yet clear what happened to Fastly. A company spokesperson said in a statement: “We have discovered a service configuration that triggered our POP interruption globally and has disabled the configuration.” “Our global network is coming back online.”
“Service configuration” can refer to any number of things; the only certainty is that no matter what the root cause is, it will have wide-ranging effects. According to Fastly’s incident report page, every continent except Antarctica has felt the impact. Even after Fastly fixed the underlying problem, it warned that users can still see a lower “cache hit rate”—how often you can find content already stored in nearby servers—and “increased raw load “, it refers to the process of returning to the source for items that are not in the cache. In other words, the cabinets are still quite empty.
Considering that CDNs are usually designed to withstand these storms, the outages are surprising. “In principle, there is a lot of redundancy,” Sitaraman said of the CDN. “If one server fails, other servers may take over the load. If the entire data center fails, the load can be transferred to other data centers. If everything is normal, you may encounter many network outages, data center problems, and server failures; The CDN’s elastic mechanism will ensure that users will never see a downgrade.”
Sitaraman said that when a problem occurs, it is usually related to a software error or configuration error, which will be pushed to multiple servers at the same time.
Even so, sites and services that use CDN usually have their own redundancy. Or at least, they should. Medina said that in fact, this morning you can see signs of the diversity of services in terms of response speed. Amazon needs about 20 minutes to resume operations because it can divert traffic to other CDN providers. Anyone who completely relies on Fastly, or who does not have the proper automation system to adapt to the interruption, will have to wait.
“The outage is the result of a single culture,” said Roland Dobbins, chief engineer at the security company Netscout Arbor. He suggested that every organization with a large amount of online business should have multiple CDN providers to avoid this situation.
However, their options are becoming more and more limited. Just as the cloud has been largely swallowed by Amazon, Google, and Microsoft, three CDN providers—Cloudflare, Akamai, and Fastly—dominate the flow of online content. “In a very small number of service providers, usage is very concentrated,” Medina said. “Whenever there is a problem with any of these three providers, it usually does not last long, but it has a significant impact on the entire Internet.”
Medina said this is an important reason why these types of blackouts have become more frequent recently, and why they will only continue to get worse. Baseball needs a defensive player; crossroads need traffic police. The fewer people you rely on, the more connections you miss and the greater the chance of a crash.
More exciting connection stories
[ad_2]
Source link
