Add a new k8s node pool dedicated to indexing jobs (with high memory VMs) with autoscaling enabled and size=0 by default to reduce cost when no indexing jobs are active.
Change VM type of graphrag api server node pool to leverage cheaper VMs that are simultaneously sufficient for query compute requirements.
Update architecture diagram to represent the new simplified networking architecture
size=0by default to reduce cost when no indexing jobs are active.