re:Invent 2022 | ✉️ #7
Hey! 👋
re:Invent 2022, the biggest AWS event of the year, is over. Frankly speaking, I did not get to watching the keynotes yet. My primary interest are the new service and feature announcements. AWS releases something new almost every single day, but re:Invent is the day when some of the biggest announcements are expected. After reading blog posts and going through the list of all the new things from the event, these are the releases that I find many AWS customers will benefit from:
AWS CloudWatch finally has a cross-account support. The key word here is “finally”, because it always felt like something that should have been there since ages. You can now collect all your metrics, logs and traces into a dedicated account, instead of constantly switching between multiple accounts
AWS RDS gets blue-green deployments. With a click, you can provision a blue/green cluster, verify some upgrades there, and easily switch to this new cluster from the old one. I remember when RDS was announced in 2013, it was just a more expensive, but primarily unhelpful way to run a database. Today, it sometimes feels like magic.
AWS Application Composer. The way to manage the infrastructure is code. But for beginners, for simple cases and for learning, this new visual tool should be great.
Lambda SnapStart. There are many things not clear about this one, but it seems like it can get rid of the cold starts entirely, only for Java functions for now.
EventBridge Pipes. AWS today has many different ways to get data into a stream, process this stream and send it further. Often, it requires connecting many services together with some custom automations. EventBridge Pipes aims at making this simpler, and simpler is always appreciated when it comes to AWS.
Of course, there were many more announcements. Most of them are rather extensions of existing services, and most of them could probably be announced any other day than re:Invent. There are some new services too: Omics, SimSpace Weaver, SupplyChain, all of them sound great, but also extremely niche, not unlike AWS Ground Station, the service to control satellites from AWS (released few years back). Would be great to try those out in a real, production environments. Some others, like VPC Lattice, I am still trying to wrap my head around.
Some of the things I hoped for did not happen. AWS EKS Fargate is still without Spot support. AWS AppRunner is still not available in Frankfurt. And RDS UI dashboards are still broken.
In general, I am glad that this year, AWS did not drown us in dozens of new services and big announcements. People are still catching up with everything released in the last 3-4 years, and re-focusing a bit on maturity of existing tools and improving them instead is a good step.
Did you keep an eye on this re:Invent? What were the things you found most fascinating, and what new services or features are you looking forward to trying out yourself?
What we've shared
What's inside the container image? Deep dive, with OCI spec as a roadmap.
You don't need a container image to run a container.
What is a Service Mesh? Figuring out what is a service mesh and why you might - or might not - need it, without looking at any particular technologies.
What we've discovered
Why you should keep using CPU Limits in Kubernetes: Answer to another article that suggests dropping CPU limits, this one explores why you should actually keep them.
Infrastructure is architecture: While I totally agree that infrastructure and architecture are indistinguishable in the cloud world, I still can't stomach that CDK is simple a YAML generator under the hood. But, replace CFN CDK with TF CDK, and things start to look a bit nicer!
Who should write Terraform code?: Lengthy exploration on where the responsibility for infrastructure code is located, taking into account the history of devops and the capabilities of your particular organization.
Delivery Lead Time in practice: How much time does it take between commit is made and commit is deployed to production? And what time is considered good? This article explores this metric on practice.
Dealing with significant Postgres database bloat — what are your options?: Deep dive into investigating the bloat of your database. Could be in particular useful for people optimising AWS Aurora RDS usage.
A correction
We know that title of dispatch #6 was the same as dispatch #5! It should have read "Not Part of Your Contract" :)
The 8th mkdev dispatch will arrive on Friday, December 23d. See you next time!