Not an April fools joke, but AppRunner is being su...
# aws
c
Not an April fools joke, but AppRunner is being sunset for new clients after April 30 2026 - https://docs.aws.amazon.com/apprunner/latest/dg/apprunner-availability-change.html. Was anyone using it here? Feels like I was the only one, but it was quite nice and cheap!
๐Ÿ‘€ 1
m
Just saw your LinkedIn post and ran here to see if anyone had shared the news. ๐Ÿ˜… I have toyed with AppRunner. For me it was a toss up because my use cases at the time required more integration (VPC, databases, and others that I canโ€™t recall) that I was able to secure with ECS and/or code it up easier with Terraform. That might have changed since AppRunner first came out. Second thing that kept me off AppRunner was slow deployments. Seemed like it took forever to reach a steady, stable state. Again, it might be faster now but for those use cases i switched to Lambda either with pure code (no container). and lastly, if i just needed to run a container, i used Google Cloud Run. It was so much faster and scaled to 0. I havenโ€™t looked at AppRunner that closely and it might scale to 0 now (maybe it always has?) but I recall it being much easier and cheaper to run a throw-away container on Cloud Run vs AppRunner. Hoping to discuss in Office Hours today!
After reading about ECS Express Mode, the ALB requirement seems like overkill. Surely that adds to the cost compared to AppRunner which was directly accessible via URL? ๐Ÿค”
c
Yep definitely. And yes AppRunner somewhat scaled to zero since you only paid for CPU resources while it was receiving requests (but you always had one active container that could run in the background!), which made it the cheapest to run long-lived tasks on AWS, even cheaper than a t4g.nano if I'm not mistaken
๐Ÿ‘๐Ÿพ 1
Deployment times were definitely slow, but for my use case was acceptable.
๐Ÿ‘๐Ÿพ 1
๐Ÿ‘ 1
m
Got it. Most of my apps at the time were almost exclusively event driven so Lambda was a better fit. For anything running longer we used EC2s (but we scaled manually vs autoscaling ๐Ÿ˜ญ). ๐Ÿ˜