2.19.6 System configuration example that uses the multi-node function
The following figure shows a system configuration example that uses the multi-node function.
|
|
![[Figure]](GRAPHICS/F0000109.GIF)
Explanation
-
Program products required on each node
You must install HADB server and HA Monitor on each node.
-
Required network
The multi-node function uses the three networks described below. These networks must be physically separated from one another.
-
Client-server network
This network is used for communication between HADB clients and HADB servers.
-
Inter-node network
This network is used for communication between HADB servers, and as HA Monitor's monitoring path.
-
Reset path
When HA Monitor detects a failure, this network is used for resetting the input/output path of the node on which the failure occurred. When using the host reset function of HA Monitor, a reset path is required.
-
-
Hardware required on each node
When using the host reset function, each node requires an SVP. An SVP is a failure management processor. A failure management processor is hardware that can set up system configurations and control the CPU.
For details about the monitoring path, reset path, SVPs, and failure management processors, see Required hardware in the manual HA Monitor for Linux(R) (x86).
- Important
-
You must select one of the following as the shared disk data protection method used by HA Monitor:
-
Host reset
-
SCSI reservation for shared disk
We recommend that you select host reset. When using host reset, an SVP must be installed in the server machines on which each node operates.
When using the SCSI reservation for shared disk method, neither an SVP nor a reset path are required. If SVPs are not installed in the server machines of each node, select the SCSI reservation for shared disk method instead.
For details about host reset and SCSI reservation for shared disk, see Host reset and SCSI reservation for shared disk in the manual HA Monitor for Linux(R) (x86).
-