.DataNodeand attempts to spread data across all volumes evenly.
.DataNodeand attempts to spread data across all volumes evenly. This is achieved by : 1) Calculating the average data that should be on a set of volumes grouped by the type. For example, how much data should be on each volume of SSDs on a machine. 2) Once we know the average data that is expected to be on a volume we move data from volumes with higher than average load to volumes with less than average load. 3) Disk Balancer operates against data nodes which are live and operational.
Copyright © 2018 Apache Software Foundation. All rights reserved.