We had previously tuned autoscaling rules in #18, but we should revisit our cluster metrics and determine if we can reduce our baseline number of running instances when the cluster is not under load. If we are able to safely reduce our baseline we could save some money while purchasing future RIs.
Brought up as a possibility in https://github.com/azavea/operations/issues/568#issuecomment-773582479.
We had previously tuned autoscaling rules in #18, but we should revisit our cluster metrics and determine if we can reduce our baseline number of running instances when the cluster is not under load. If we are able to safely reduce our baseline we could save some money while purchasing future RIs.
Brought up as a possibility in https://github.com/azavea/operations/issues/568#issuecomment-773582479.