The following architecture shows an ApsaraDB for Cassandra instance that contains two data centers. Each data center contains four nodes.

  • An ApsaraDB for Cassandra instance can contain one to three data centers, each data center contains one to 500 nodes that each deploys a Cassandra process.
  • ApsaraDB for Cassandra instances do not adopt the master-proxy node architecture, all nodes are peers. Logically, the system specifies two nodes in each data center as the seed nodes. However, seed nodes do not require process deployment because seed nodes bring little impact on the system.
  • The client connects to all server nodes to ensure high performance.
  • The number of replicas of each data center can be customized as needed. Replicas can be used for data center disaster recovery, backup, and offline analysis.




The storage space of a node ranges from 160 GB to 16 TB. Five hundred nodes can contain up to several PB of storage space.