Scale-to-zero
in progress
Yann Léger
To reduce costs on applications which do not have constant traffic and are not latency sensitive, a scale-to-zero feature would be useful.
Log In
Saybo podiyam
David Lechevalier
in progress
Jonas Stjernquist
I would like to have the ability to have scale-to-zero in combination with provided settings for minimum count of services which is active.
Examples of autoscale + scale-to-zero configurations:
3 x eSmall instances always running
1 - 3 x eSmall instances always running
1 x eSmall instance always running and up to 3 running eSmall instances
Satyendra
Hmm, Seems nice. It would be better to scale back up after a request comes in. :)
and also an option to scheduling turn off/on times for a service is appreciable!
Yann Léger
planned
Yann Léger
under review
Édouard Bonlieu
Chris Helder
Building off of Seth's suggestion, it would also be nice to make sure this is a toggle switch, for those who may want services running 24/7.
Kevin Bird
Is there a way to scale back up after a request comes in? I think this is a great idea but would really love a way to scale back up to 1 after a user hits it after a long downtime
Yann Léger
Kevin Bird: Yes, that's the target. The feature will scale-to-zero if there is no request for a specific amount of time and automatic scale back when requests hit the service mesh depending on the autoscaling parameters.
Seth
Perhaps an additional nice to have feature would be to only scale-to-zero during specific time slots. Although this might be a different use-case then auto scale-to-zero.
Yann Léger
Seth: Are you thinking about completely suspending services during specific time slots or only enabling scale-to-zero during specific time?
I think this is something which Scheduled jobs can accomplish, but it would indeed be even easier to setup if it was completely built-in.
Load More
→