We are running some experiments on our infrastructure testing our current resource usage. This involves reducing the amount of resources we have allocated to the cluster to measure how much we *actually* need. We aren't expecting any performance impact, but are actively monitoring. If you see any weirdness, please let us know!