Good Morning!
A handful of interesting releases, and a couple of infuriating ones await you in this week’s issue.
What do you folks want to see more / less of? Hit reply and let me know.
From the Community
A year ago Redis (a proprietary fork of Valkey) changed its license and lost most of its external contributors.
The fallout from Oracle’s attempt to hide a breach from customers continues to land.
Amazon quietly bids for TikTok hoping to ruin another attempt at being a social network.
Intercom used PlanetScale to do some really interesting things. Folks I trust there say nice things about it, so–although it’s no Route 53–I’m looking more seriously at the product…
Podcasts
Last Week In AWS: Northern Virginia is in Virginia
Screaming in the Cloud: Making Dropping and Sharing Easy with Timo Josten
Choice Cuts
Amazon EC2 now supports more bandwidth and jumbo frames to select destinations – Son of a–I just explained to a room full of people at SREcon that jumbo frames didn’t cross gateway boundaries in AWS networks. Sweet, I’ve inadvertently incorrected a whole crop of folks.
API Gateway launches support for dual-stack (IPv4 and IPv6) endpoints – This is a super handy feature that’s quite timely, assuming it’s still 2015.
AWS Lambda adds support for Ruby 3.4 – So soon?
Amazon CloudWatch Logs increases maximum log event size to 1 MB – Sweet Jesus, what the hell are you people logging? Database dumps? Images?
Amazon Neptune announces 99.99% availability Service Level Agreement – A four nines SLA is going to be super useful for the four customers who have a legitimate use case for Neptune.
Announcing the general availability of Amazon VPC Route Server – Ooh, if I learn BGP can I use it for things like alterNAT to smooth the route table transition between the NAT instance and the dread Managed NAT Gateway?
Under the hood: Amazon EKS Auto Mode – "Kubernetes as a service" I’m sorry what the hell is this? Kubernetes lets you cosplay as a cloud provider, so now an actual cloud provider is running Kubernetes for you so you can run an abstraction on top of an abstraction in order to… what, exactly?
Optimizing cost savings: The advantage of Amazon Aurora over self-managed open source databases – A sales pitch for Aurora that doesn’t mention PlanetScale, to which it loses badly along virtually every economic axis. My personal favorite: I can use existing Savings Plans to cover PlanetScale instances, but I can’t do that with Aurora because the RDS team seconds Customer Obsession to Margin Obsession.
How AWS Sales uses generative AI to streamline account planning – I wonder if this has anything to do with a phone call I got last week from AWS sales, asking about my environment at Expensify. Now yes, I did work at Expensify–but I left in 2012. That timing would however be just about perfect for someone at IBM Cloud to be inquiring after my business.
Issue with AWS SAM CLI (CVE-2025-3047, CVE-2025-3048) – Update your SAM installation, which still remains the best way I’ve found to quickly deploy Lambda functions. Sure wish that market hadn’t stalled.
They’re trying to keep it quiet, but once again the SageMaker team has moved too quickly and given too many permissions to an IAM role. Emails went out but you should know by now that you can’t hide from me, AWS.
Tools
An MCP so your GenAI thing can query CloudWatch logs on your behalf.
… and that’s what happened Last Week in AWS.
