The Cloud Pod | Weekly AI & Cloud News on AWS, Azure & GCP · Justin Brodley, Jonathan Baker, Ryan Lucas and Matt Kohn | Cloud Computing & AI News

AWS Finally Admits Kubernetes is Hard, Makes Robots Do It Instead

·1 hr 15 min·5 clips

AWS US East 1 had a major outage that hit Facebook, Snapchat, Coinbase, ChatGPT, and Amazon services.

The episode begins with the aftershocks of a rough week in cloud. Amazon had a major outage, and the hosts lean into the irony of surviving it by being on other platforms. They also revisit last week’s discussion of Glacier deprecations. An email from Amazon clarifies that the standalone Amazon Glacier service is the one being retired. From there, the conversation moves into a new CPU optimization feature for EC2. The hosts work through what it means for vCPU-based licensing costs and why the announcement feels a little odd. They compare it to custom shapes and to controls that already exist lower in the stack. The discussion turns practical fast. Rather than celebrate more knobs, they keep asking whether the better answer is simply to use a managed service. That leads into a familiar refrain about RDS and Microsoft SQL. The panel is sympathetic to the pain, but they still prefer the route that reduces manual work. The episode then shifts to patching in managed services. One host notes that this is basically day-job material. AWS Systems Manager Patch Manager now includes an available security update state for Windows, and the team unpacks why that matters operationally. The tone stays brisk and familiar. They keep puncturing vendor language with workplace reality. They also keep circling back to the same lesson: cloud features are only useful if they lower friction in production. The closing stretch lands on the same theme. Every clever workaround has a cleanup cost somewhere. The show ends where it often does, with a mix of fatigue, dry humor, and respect for anything that reduces the amount of real work humans have to do.

As heard by us

A sharp cloud-news recap that treats outages, licensing, and patching like real operational problems.

The episode starts with the AWS outage that made the internet feel half-broken, then moves through Glacier deprecation clarification, CPU tuning for licensing-heavy workloads, and a patching update that lands in day-job territory. The strongest stretch is the back-and-forth.

Read the full review in PlayNext →

Why you'd press play

You want the cloud week translated into practical consequences, not vendor theater.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from The Cloud Pod | Weekly AI & Cloud News on AWS, Azure & GCP