Hi All, Just wanted to share one activity on whic...
# general
r
Hi All, Just wanted to share one activity on which I am currently working on, it is an onboarding a new application to Azure Container Apps as an alternative to Azure Kubernetes Service (AKS). The primary motivation behind this shift is cost optimization. Unlike AKS, where each pod and node consumes IP addresses (contributing to subnet exhaustion and increased cost), Azure Container Apps helps mitigate this by handling infrastructure and networking more efficiently. Container Apps abstracts away the need for managing individual nodes or pods, leading to lower overhead in terms of IP allocation. Additionally, by reducing infrastructure complexity, we can improve the performance and reduced latency, which should contribute to a smoother runtime experience for the application.
b
Isn't it like comparing apples to oranges? Does ACA offer the same configurability as AKS does? However, if the migration fits your needs and fixes your budget problems, then I'm happy for the move. 👍
r
Absolutely not, the app which we are onboarding is something that can run smoothly in both aks or container apps. Also I am featuring out which apps can be migrated from aks to aca without causing issues
n
that’s also not an accurate reading of the landscape regarding costs. Using older CNIs, that’s true, but the modern options do not typically allocate one IP/pod any more
But you could save money in some cases, at a trade of reduced flexibility
j
Most web service / workers can run in a container-scheduler (with or without a loadbalancer). But if you run into situations where k8s operators, databases, non-native Azure services are necessary, then you obviously would want AKS. Same applies for AWS Fargate vs EKS, or Google AppEngine vs GKE