Node pools enable multiple types of differnent computing capabilites to be utilized within a single cluster. it weould be great to have this as we are looking to leverage more compute intensive applications with stateful datasets and being able to delegate workloads using taints and tolerations would help not only with perfomance but costs as well.