- Adding more data nodes to the cluster
- Increasing the number of replicas
- Upgrading the instance types of the data nodes
- Reducing the number of shards