Replication

To provide fault-tolerance and prevent loss of tuples if one of the seeders in a space suddenly goes down, you can specify that space data is replicated—backed up one or more seeders.

Redistribution or Replication

  Redistribution is performed when a new seeder joins the space, rereplication happens when a seeder leaves the space.

Distributed Replication

 Replication is distributed over several seeders. Each seeder seeds some tuples and also replicates tuples assigned to other seeders. A given seeder does not have a designated backup that replicates all of the tuples that this seeder seeds; instead, the tuples that it seeds are replicated by any of the other seeders.

Active-active Mode

 There are no “backup seeders” waiting for a “primary seeder” to fail to start before becoming active. Instead, all seeders are always active, and are both seeding and replicating tuples. This ensures efficient replication— if a seeder fails, there is no need to redistribute the tuples among the remaining seeders to ensure that the distribution remains balanced. ActiveSpaces simply rebuilds the replication data, which is less work than having to redistribute. This results in a lower performance impact when a seeder fails.

• Degrees of Replication
  When you configure a space you specify whether replication is enabled, and if it is enabled, specify the replication value using the SpaceDef object’s setReplicationCount or getReplicationCount methods. The replication value specifies whether replication is enabled, and if it is enabled, how may seeders participate in replication.
• Phase Count and Phase Ratio Tuning Redistribution and Replication
  You can control the pace and throttling of redistribution and replication be by using the space definition settings.
• Synchronous and Asynchronous Replication
  You can also define whether replication is performed in synchronous or asynchronous mode for the space.