I have an Elasticsearch cluster with three data nodes. For some experiment, I want to move all the shards to a single data node. So I used the shard allocation filter feature to achieve that. From Marvel, I see that that Network Transport Bytes Sent speed is around 16MBps each from the two source data nodes and 32MBps on the target data node. I want to know if there is an Elasticsearch configuration to increase the speed to moving shards across data nodes.
This can be managed via settings described here https://www.elastic.co/guide/en/elasticsearch/reference/current/modules-cluster.html
Specifically you could increase cluster.routing.allocation.cluster_concurrent_rebalance.