AWS shock: The world needs a more reliable internet

Last Monday, the internet’s digital bloodstream developed a clot that sent jitters around the world. Amazon Web Services (AWS) began reporting increased error rates and latency in its flagship cloud region, US-East-1, in Northern Virginia.

- SIDDHARTH PAI

The cause was eventually traced to a Domain Name System (DNS) resolution failure affecting a critical application programming interface (API) endpoint for DynamoDB, AWS’s popular server-less database service.

This DNS issue meant that software trying to connect to DynamoDB couldn't find it; not because the database was offline, but the system responsible for translating human-readable names into machine-usable Internet Protocol (IP) addresses had stopped functioning properly. Application requests failed, retries spiked and global services got disrupted, hitting everything from apps such as Snapchat and Signal to government services. To understand the implications, it helps to look at how DNS works. At its core, DNS is the internet's phone book. When you enter a web address like ‘amazon.com,’ DNS translates that into a numeric IP address so your browser can locate the correct server. This involves a chain of lookups: from the root servers to top-level domain servers, such as those for a *.com’ address, to the authoritative server for the specific domain.