Steel-Belted Radius Carrier 7.2.x Installaion Guide > Planning Your Session State Register Cluster > Planning the Cluster Topology

Planning the Cluster Topology

The topology of all clusters begins with the four machines required to implement the SSR Starter Kit: two machines that each host both an SBR node and an SSR management node and two machines that each host one SSR data node.

Figure 9: Basic Steel-Belted Radius Carrier Session State Register Starter Kit Cluster

Naming the Cluster and Its Machines

We recommend that you follow a machine naming convention that uses the cluster name as a building block to construct each machine's system name and hostnames, but this is not required. Table 7 shows the recommended syntax to build the machine names.

Table 7: Cluster Hostnaming Syntax

Cluster Name	System Name and Hostname
<cluster name>	<cluster name>-<machine number>

To determine the machine names:

1. Name the cluster.

The cluster name must:

• Begin with a letter (an alphabetic character)
• Be made up of alphanumeric characters, underbars, and dashes.
• Be a single string without any spaces

2. Create the system name — Add a dash and a machine number to the cluster name. The Solaris convention for naming is to set the system name to be the same as the hostname associated with the IP address of the primary network interface. You can test the name with the uname -n command.

The machine number may be any one-, two-, or three-digit number.

It is not required, but you can simplify machine identification if you adopt a convention that builds on the node identification numbers (node ID) that are used by the SSR processes and incorporate them into the server name.

The general scheme for node IDs is shown in Table 8.

Table 8: Allocation of Node IDs

Cluster	Single*	Purpose
0	0	Internal use
1,2,3	4	Management processes (ndb_mgmd) on SBR/management combination (sm) nodes and management (m) nodes
6,7,8	5	mysqld processes (mysqld) on SBR/management combination (sm) nodes and management (m) nodes
10-29	30-36	Data processes (ndbd) on data (d) nodes
30-39	37-39	Reserved
40-59	9	SBR processes (radiusd) on SBR (s) nodes
60-63	60-63	Spare

* Reserved for stand-alone and transition servers.

For example, to put this into practice, you could assign the Starter Kit SBR/management (sm) nodes names and IP addresses ending in the range 1-3 (or 6-8 because 1-3 are often associated with routers), and you could assign the Starter Kit data (d) nodes names and IP addresses ending in the range 10-29.

Starter Kit Cluster Naming Example — the Blue Cluster

Table 9 shows how machine and interface names are built from the cluster name using the node ID mentioned in Step 2 on 41. This example for a SSR Starter Kit creates the blue cluster. The cluster and node entries are used in subsequent chapters to create a working cluster example. Because multipathing is required, only one IP address is used for each machine; in this example, because the multipath addresses are assigned, each server's IP address is set to echo each server's primary node ID number.

Table 9: Example Blue Cluster Naming Worksheet 

Product Name	Cluster Name	Hosted Nodes	System Name and Hostname	Multipath IP Address	Node ID Numbers
	<cluster name>		<cluster name>—<machine no.>
SSR Starter Kit	blue	SBR and Mgt.	blue—1	192.168.0.1	ndb_mgmd =1
					mysqld =6
					radiusd =41
		SBR and Mgt.	blue—2	192.168.0.2	ndb_mgmd =2
					mysqld =7
					radiusd =42
		Data	blue—18	192.168.0.18	ndbd =18
		Data	blue—19	192.168.0.19	ndbd =19

Implementing the worksheet in Table 9 results in the cluster setup shown in Figure 10.

Figure 10: The Blue Starter Kit Cluster