Super Node: Super Node is based on an active-active (high-availability) architecture with read-write separation. EAZStor metadata is divided into libraries and tables and stored in Super Nodes across regions. Each library and table adopts a Primary-replica model on different nodes to form an overall active-active architecture. Furthermore each Super Node itself also adopts a Primary-replica model to construct a small cluster with active-active characteristics. EAZStor system network supports up to 1024 Super Nodes.

Data Node: Data Node is based on a completely decentralized peer-to-peer storage architecture which consists of a mass number of Data Nodes forming a large-scale storage cluster. By encrypting user data and distributing it on different Data Nodes across regions, and mix with high fault
tolerance erasure codes, Data Node ensures the availability of user data.

User Node User Node (also known as Edge Node) runs on user system and provides application interfaces for user management, key management, and S3 backend services; as well as execution of
compression, block segmentation, encryption, deduplication, and erasure correction code
calculation.

Relay Server: If Super Node, Data Node, and User Node are located behind routers with private IP address, they cannot communicate directly through IP address. In this case, a Relay Server is needed to assist in connection and establish P2P communication. The Relay Server has a dedicated public IP address, and all nodes can communicate directly with it. It marks each node by node ID, maintains a node mapping table internally, receives P2P connection requests from clients, locates the communicating parties according to the node ID in the request, and notifies and assists in establishing
the connection. .