Rachel Mak, Photography Editor

On the first day after October break, students logging onto Canvas, Yale’s online course management system, were met with a message that Canvas was “experiencing issues due to an ongoing AWS incident,” referring to a widespread Amazon Web Services outage.

Early Monday morning, disruptions to Amazon Web Services, or AWS, a cloud service provider, led to outages for hundreds of websites, including Canvas, the learning management system used by Yale. Access to Canvas was restored Monday evening.

“I am encouraging professors to use their flexibility and ingenuity in responding to this unusual situation; if the outage goes on much longer, we will issue guidance, but it looks like it is already being resolved globally,” Yale College Dean Pericles Lewis wrote in an email to the News Monday afternoon before Canvas was restored.

Some students felt that the Canvas shutdown was especially impactful this time of year, when students often have midterm exams.

“The canvas shutdown has me tweaking out—I’m a girl that religiously checks her canvas so not being able to submit my assignments or even see them has been really worrisome,” Faven Wondwosen ’27 wrote to the News.

Wondwosen added that she relies on her online assignments and the materials her teachers have uploaded to Canvas to help her study for her exams.

Another student, Nikolai Stephens-Zumbaum ’26, said that the Canvas outage was helpful for some students whose exams or assignments were postponed because they couldn’t access necessary course materials.

Anurag Khandelwal, an assistant professor of computer science who is teaching, told the News that the AWS outage seems to have occurred in two phases.

Khandelwal, referring to outage updates by AWS, said that the initial failure was likely related to the Domain Name System, or DNS, which associates domain names — such as yale.edu — with numeric addresses. DNS is needed for websites and services such as Canvas to access DynamoDB, a database service provided by AWS, Khandelwal wrote in an email.

By early morning on Monday, an update by AWS wrote that the “underlying DNS issue was fully mitigated at 2:24 AM PDT.” However, the outage continued, and another update at 7:14 a.m. PDT confirmed “significant API errors and connectivity issues.”

Khandelwal wrote that the previous outage may have caused another outage if AWS was overwhelmed by a backlog of queries.

“Once the DNS failure was resolved, there was still likely a backlog of query requests that had to be serviced by the DynamoDB after recovery. This backlog — combined with new queries coming in — can overwhelm the system,” Khandelwal wrote. 

By Monday evening, the issue was marked as “resolved” on AWS’ status website, and the latest Monday update said that backlogs would be processed “over the next few hours.”

According to an announcement by AWS, the outage was resolved at 2:24 a.m. PDT on Monday. Instructure, the Utah-based company that operates Canvas, announced that Canvas became accessible at 5:16 p.m. MDT.

Brian Watkins, an Instructure spokesperson, wrote in an email to the News on Monday evening at 8:01 p.m. confirming that Canvas access had been restored.

“We recognize the integral role Canvas plays in the daily lives of educators and students, serving as a central hub for teaching and learning, and we know today’s AWS outage had a significant impact on that experience,” Watkins wrote. “Our teams have worked closely with Amazon Web Services (AWS) to restore service and continue to monitor stability. All affected users should now have full access.”

Khandelwal wrote that the usage of cloud computing platforms such as AWS, Google Cloud Platform or Microsoft Azure is mainly due to “cost-efficiency, flexibility and ease of deployment,” akin to the difference between renting and purchasing a home.

“Canvas could host its service on its own infrastructure (at significant capital investment), but this would not necessarily avoid such outages — it might actually make it worse since they may not have enough resources to stay on top of issues that arise,” Khandelwal wrote. 

AWS reported an operating income of $39.8 billion for 2024.

Olivia Woo contributed reporting.

Update, Oct. 21: This article has been updated to include a statement from Brian Watkins, an Instructure spokesperson.