Two-Node Cluster

The quickest setup for a two-node cluster is to use rackmounted servers. Whichever rackmounted server you use, the cluster must be configured as follows (with IP replication used for data sharing between the two master-eligible nodes):

• Two rackmounted servers configured as master-eligible nodes.

• Some partitions of the rackmounted server’s internal disks configured as replicated partitions (for storage of highly available data).

• On-board Gigabit Ethernet interfaces (or supplementary cards) of the two rack-mounted servers for the configuration of the cluster network and (optionally) for providing external access to the cluster. In the latter case, use at least four NICs.

• Two external Ethernet switches (see Chapter 1 of the Netra High Availability Suite 3.0 1/08 Foundation Services Installation Guide for information about configuring and connecting the switches).

• One terminal server to manage the consoles (see Chapter 1 of the Netra High Availability Suite 3.0 1/08 Foundation Services Installation Guide for information about configuring and connecting the terminal server).

The following list provides some examples of a two-node cluster based on rack-mounted servers:

• Two-node cluster with two Netra 120 or two SunFire V210 servers

• Two-node cluster with two Netra 240 or two SunFire V240 servers

• Two-node cluster with two Netra 440 or two SunFire V440 servers

• Two-node cluster with two Netra T2000 servers

Choose Netra T2000 servers if you want to test the behavior and performance of the Netra HA Suite Foundation Services on CMT-based hardware.

---

Note - If you need to evaluate data sharing between two MENs by using shared disks (instead of IP replication), add an external disk array to your configuration. Check with your support team to determine which external disks are supported for use on a particular server.

---

---

Note - All of the servers listed in this section are running with only the Solaris OS. To evaluate the Foundations Services on Linux, you must set up a cluster with two Netra CP3020 blades (Opteron-based blades) in an ATCA chassis (preferably the Sun Netra CT900 chassis).

---

Four-Node Cluster

If you already have a two-node cluster set up as described in the preceding section, and want to evaluate the testing of client nodes, you can either add new rack-mounted servers connected to the cluster network, or add an ATCA chassis with two blades to your configuration. The second option is recommended, as it offers more flexibility in the choice of which client nodes you use. For example, you can use either SPARC®, Opteron™, or CMT-based blades running the Solaris OS or Linux, and will then have the potential to test scalability afterwards (you can easily add up to 12 blades in the chassis).A four-node cluster based on two rackmounted servers and two ATCA blades must contain the following:

• Two rackmounted servers configured as master-eligible nodes.

• One ATCA (Netra CT 900) chassis and two ATCA blades configured as master-ineligible nodes (diskless or dataless).

• Replicated partitions defined on the internal disks of the rackmounted servers (for storage of highly available data).

• On-board Gigabit Ethernet interfaces (or supplementary cards) of the rack-mounted servers for the configuration of the cluster network and (optionally) for providing external access to the cluster. In the latter case, use of at least four NICs is recommended.

• Two external Ethernet switches (see Chapter 1 of the Netra High Availability Suite 3.0 1/08 Foundation Services Installation Guide for information about configuring and connecting the switches).

• One terminal server to manage the consoles (see Chapter 1 of the Netra High Availability Suite 3.0 1/08 Foundation Services Installation Guide for information about configuring and connecting the terminal server).

The following are examples of a four-node cluster based on two rackmounted ATCA blades:

• Four-node cluster with two Netra 240 servers and one Netra CT 900 chassis containing two Netra CP3010 or two Netra CP 3020 blades.

• Four-node cluster with two Netra 440 servers and one Netra CT 900 chassis containing two Netra CP3010 or two Netra CP 3020 blades.

• Four-node cluster with two Netra T2000 servers and one Netra CT 900 chassis containing two Netra CP3060 blades.

---

Note - If you need to evaluate data sharing between two MENs by using shared disks (instead of IP replication), add an external disk array to your configuration. Check with your support team to determine which external disks are supported for use on a particular server.

---

---

Note - If you intend for your cluster to have all nodes running Linux, to evaluate the Foundations Services, you must set up a cluster with only Netra CP3020 blades (Opteron-based blades) in one or two ATCA chassis (preferably the Sun Netra CT900 chassis).

---

If you are installing a four-node cluster from scratch, use an ATCA blade server (Netra CT900 chassis with Netra CP30xx blades) to ensure maximum flexibility and possible scalability. A four-node cluster based on an ATCA blade server must be configured with the following hardware:

• One ATCA (Netra CT 900) chassis in which the four blades are in the same chassis, or two ATCA chassis in which two blades (one MEN and one NMEN) are in each chassis.

• Two ATCA blades (Netra CP30xx) configured as master-eligible nodes.

• Two ATCA blades (Netra CP30xx) configured as master-ineligible nodes (diskless or dataless).

• Replicated partitions defined on the internal disks of the two MENs (for storage of highly available data).

• Internal ATCA base or extended fabrics used for the cluster network configuration. Other Gigabit Ethernet interfaces (available on the blades) are used for connecting external networks (if required).

• Two ATCA switch blades (see Chapter 1 of the Netra High Availability Suite 3.0 1/08 Foundation Services Installation Guide for information about configuring and connecting the switches).

• Two ATCA shelf manager blades to manage the consoles (see Chapter 1 of the Netra High Availability Suite 3.0 1/08 Foundation Services Installation Guide for information about configuring and connecting the terminal server).

The following are examples of a four-node cluster based on an ATCA blade server:

• Four-node cluster with one Netra CT 900 chassis containing four Netra CP3010, four Netra CP3020, or four Netra CP3060 blades.

• Four- node cluster with one Netra CT 900 chassis containing two Netra CP3010 blades and two Netra CP3060 blades, or two Netra CP3020 blades and two Netra CP3060 blades.

• Four-node cluster with two Netra CT 900 chassis containing two Netra CP3020 or two Netra CP3060 blades in each chassis (one MEN and one NMEN in each chassis).

---

Note - If you plan to evaluate a four-node cluster on a system where Logical Domains are in use, build your cluster using either an ATCA blade server with two Netra CP3060 blades or two Netra rackmounted T2000 servers (both are UltraSPARC T1 processor-based hardware). On each Netra CP 3060 blade (or each Netra T2000 server) you will configure three domains (one control domain and two guest domains). Each domain will run the Solaris OS. Each guest domain will be considered as a node of the cluster. There will be one master-eligible and one master-ineligible node on each Netra CP3060 blade or Netra T2000 server.

---