Scaling the cluster
As workloads in the cluster increase, you must add more CPU, memory, GPU, or storage to accommodate them. CPU and memory can be increased by adding additional compute nodes to the cluster. For workloads that require GPUs, you can add specialized GPU nodes to the cluster. Storage can be scaled by adding additional disks to existing storage nodes, or adding additional storage nodes to the cluster. If you want to scale your cluster beyond a single rack, then you can add up to two expansion racks, enabling your cluster to host more compute, storage, and GPU nodes.