The Silver Lining

Cloud outages are a pain, but each one dispenses valuable lessons

We’ve all experienced it. You log in to discover that various parts of the internet – included key sites and services you need to go about your business – are down. “How long with the outage last?” you wonder as you scramble to rearrange your day. It’s not only frustrating, it’s expensive. Depending on the size of your company, cloud outages can cost thousands or even millions of dollars per minute.

In a weird way, these outages can have a positive effect on cloud providers, as people see just how many of the biggest apps rely on them.

“When AWS went down, their share price went up, because people realized how many people are using them,” University College London computer science professor Steven Murdoch told The Guardian. “In some ways (the outage) is great marketing, because you see how many people are using Cloudflare.”

But headline-making outages are not something cloud providers want to become known for or something customers want to deal with regularly. Here’s how providers are working to combat the issue and what businesses can do to minimize the effects of a cloud outage.

Caption of Image

Time Is Money

That AWS outage on Oct. 20 of last year in its us-east-1 region cost U.S. companies approximately $500-$650 million, according to insurer Parametrix. The total outage time was 2 hours, 36 minutes. That’s an eye-popping testimonial to how much money flows through the U.S. economy on a given day and how much of that commerce relies on cloud internet.

Just a few weeks later, AWS and Google Cloud announced a joint multicloud networking service that combines AWS’ Interconnect-multicloud with Google Cloud’s Cross-Cloud Interconnect.

“… (C)ustomers no longer need to worry about any heavy lifting to create their desired connectivity. When they need multicloud connectivity, it’s ready to activate in minutes with a simple point and click,” Robert Kennedy, VP of network services for AWS, said.

The partnership enables customers to move data and applications between clouds quickly and easily. Provided employees have the tools and information needed on both, they could switch to the other when one cloud is down, then import any new data when both are back up.

“This profound simplification enables end-to-end connectivity in minutes, transforming multicloud deployment from a days-long engineering project into a simple, rapid configuration task,” a Google Cloud blog post stated.

Don’t Panic

The centralized nature of cloud networks might leave them vulnerable to outages, but Gartner warns that businesses moving their data and operations to smaller, more localized clouds won’t eliminate the risk of outages and might lead to even longer recovery times when networks do go down. The big providers might have outages, but they’re motivated and equipped to get things up and running again as fast as possible.

“These events highlight an important truth: cloud disruptions happen, but they are not evidence that the cloud is inherently unreliable,” Gartner analyst Lydia Leong wrote. “In fact, these incidents follow a familiar pattern seen across all hyperscale cloud providers over the last decade: a regional event, typically resolved within hours, often tied to a network dependency, but not affecting running compute instances themselves.”

What Gartner recommends is that, knowing that it’s a matter of when not if cloud outages will occur, app developers spread out workloads over multiple availability zones. If one region’s coverage goes out, they can fail over quickly to another. It’s quite a bit like the AWS-Google multicloud solution, but Gartner suggests that “investing in robust architectures within one cloud delivers better uptime and simpler operations than trying to juggle multiple providers with different APIs and processes.”

What enterprises should do instead is focus on substitutability, having manual workarounds or alternative platforms they can turn to in case of an outage without spending the money necessary to be on multiple clouds.

Agents Are On It

AWS’ cloud unit has been busy working on other outage solutions as well. In December, AWS unveiled DevOps Agent, and AI tool that helps businesses quickly discern the source of a network outage and potential fixes. With assistance from Datadog and Dynatrace, DevOps Agent researches different potential causes of an outage without waiting for human site reliability engineers to start.

In a trial with Commonwealth Bank of Australia, DevOps Agent took 15 minutes to diagnose an outage cause that would have taken an engineer hours to discover, AWS said.

“By the time the on-call ops team member dials in, they have an incident report with preliminary investigation of what could be the likely outcome and then suggest what could be the remediation as well,” Swami Sivasubramanian, VP of agentic AI at AWS, told CNBC.

As agentic AI continues to improve and scale, providers might deploy them to their own regional cloud outages, drastically cutting downtime and saving millions. After all, time is money.