A company uses an Amazon Redshift provisioned cluster as its database service and the cluster uses key distribution. While running the maintenance tasks on the cluster, a data engineer noticed that of the four ra3 cluster nodes, one node had a CPU load of over 90% most of the time, while the other three nodes were averaging around 10%. The company has tasked the data engineer to balance the load more evenly across all four compute nodes. What changes are needed if provisioning additional nodes is not an option?