NCache 4.6 - Online Documentation

Data Load Balancing in Partitioned and Partitioned-Replica Topologies

Data distribution in partitioned and partitioned-replicated cluster caches is based on hash code of cache keys. Data is almost evenly distributed among all server nodes.
Each node knows about the data load of every other node in the cluster. There can be scenarios when data load on any one node becomes high depending upon the pattern of cache keys inserted in the cache.
To cater with uneven loading, NCache provides you with load balancing options in following ways:
Auto load balancing
  • Configurable: Auto load balancing is a configurable feature which automatically balances data load on each and every server node.
  • Automatic: It continuously monitors data load on all servers of the cluster after a configurable interval and detects uneven data load.
  • Threshold: It verifies load against a configurable threshold, if any of the cache node has imbalanced data then coordinator generates new distribution maps to redistribute the extra data from that node.
  • Turning off: You can also turn off this feature if you do not want load balancing among cluster nodes.
Manual Load balancing
  • Manual: You can also balance data load any time by a manual load balancing feature through NCache Manager.
  • Balance data on overloaded server:  You can trigger data balancing through selecting an option on the server node where you think data load is more than other server nodes.
  • No threshold: There is no threshold; you can start data balancing when ever required.
See Also