3a. Basic principles, terminology

Bagri is a client-server system. The server side is built on top of distributed cache (Hazelcast in this instance) and can be deployed as a number of data servers (cache nodes). Every started data server instance manages data for one or more schemas. Schemas are are like database schemas in RDBMS world. Several data servers managing data for the same schema forms data cluster.

Clients can connect to the server in two ways:

• via standard Bagri XQJ driver which implements JSR-225 interfaces (the same as JDBC driver for a database);
• via proprietary Bagri XDM API, which is used by the XQJ driver and provides some additional functionality not covered by JSR-225 spec;

A set of Bagri Data servers and schemas deployed on them are managed by Admin server. The Admin server is also used for management of access and other common server artifacts. It is used for administration and monitoring purpose only and is not strictly required for the system to work properly.