2.19.1 About the multi-node function

The multi-node function realizes the load distribution of retrieval SQL statement processing by coordinating multiple HADB servers (scale-out based load distribution).

The following figure shows an example of using the multi-node function to distribute the processing load.

Figure 2‒68: Example of using the multi-node function to distribute the processing load

Explanation

In the example above, three HADB servers are used to execute the retrieval SQL statement, enabling load distribution.

1. Application program connection requests from HADB clients are received by the HADB server on the master node.

2. The HADB server on the master node controls load distribution. It does this by selecting a node with a lighter load, and executing retrieval SQL statements on that node. Load distribution is achieved by allocating processing of retrieval SQL statements to the individual nodes.

• Master node

  The node containing the HADB server that controls the multi-node function is called the master node. Update SQL statements and commands such as the adbimport command are processed on the master node.

  One master node is required.

  Application program or command connection requests are received by the HADB server on the master node.

• Slave node

  A node containing an HADB server that processes retrieval SQL statements is called a slave node. When the multi-node function is used, the maximum number of nodes is four. Therefore, you can set up between one and three slave nodes.

  Update SQL statements and some commands, such as the adbimport command, cannot be executed on a slave node.

• Multi-node configuration

  A system configuration consisting of multiple HADB servers that are using the multi-node function is called a multi-node configuration.

• HA Monitor

  This program product monitors node statuses and separates any node on which a failure occurs. For details about the functionality of HA Monitor, see 2.19.3 Managing nodes with HA Monitor.

  The master node operates as the active system of HA Monitor, and the slave nodes operate as the standby systems of HA Monitor.

Note

A server machine on which an HADB server is installed, including the operating system, HADB server, and HA Monitor installed on that server machine, is referred to as a node.

Processing performance when the multi-node function is used depends on the hardware environment and the amount of data being processed. Note that using the multi-node function might increase the time required to execute SQL statements and commands because the overhead for communication or synchronization between the master and slave nodes increases.

To Page Top