Interesting, thanks for the updates.
I’d like to bring up a point that I’m not sure is strictly related to DNS. I imagine it’s more related to the load balancer (unless there’s some possibility of doing this behavior via DNS).
Here’s the scenario:
-
Imagine I have an application with 2 machines to balance load/traffic.
-
Let’s say their soft limit is configured to 5 requests.
-
The application is below the soft limit, but there are still 2 machines active.
-
In that case, why not balance traffic between the two active machines instead of redirecting all traffic to only one?
What I observe is that when I have two machines running, even if the load is below the soft limit, all traffic is routed to the same machine, and the second one only starts receiving traffic when concurrency increases.
I would expect something closer to a round-robin behavior, distributing requests between the active machines, even if it is below the soft limit.
PS: I know about auto-stop machines, that’s not the point here. I’m wondering whether something like this is on the roadmap to be implemented in the future.