Support for granular downscaling #263

merged 3 commits into from Sep 30, 2015


None yet

2 participants


Much like #258, add granular downscaling as well. Only supports downscaling on consumed vs provisioned capacity, as I can't imagine a situation where one would want to scale DOWN when they have throttling.

With this kind of setup a user could, for example, only scale down slightly when at a certain point, but also immediately scale all the way down to 1 read/write unit when they have 0% utilization. This would prevent a situation like described in #245 where the autoscaler would only downscale very few units each iteration, risking bumping up against the downscale limit.

I'll work on the docs if this looks good to you.

On a side note, I think a lot of this code could be consolidated together, as I performed more copy/paste of large chunks of code than I'd care to admit, but I figured that refactoring would be out of the scope of this feature. I'd like to visit it after this and my other PRs are merged, though.

@sebdah sebdah self-assigned this Sep 11, 2015
@sebdah sebdah added this to the 1.20.x milestone Sep 11, 2015
@sebdah sebdah modified the milestone: 1.21.x, 1.20.x Sep 11, 2015
sebdah commented Sep 11, 2015

Thanks for the PR. I'm gonna look it through and likely include it in the 2.1.0 release.

@sebdah sebdah added this to the 2.1.x milestone Sep 11, 2015
@sebdah sebdah added a commit that referenced this pull request Sep 30, 2015
@sebdah Add release notes for #263 36a4e1b
@sebdah sebdah merged commit 07d23db into sebdah:master Sep 30, 2015
sebdah commented Sep 30, 2015

This has now been released in version 2.1.0! Thanks again for the PR @Sazpaimon

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment