Understanding and Designing Serviceguard Disaster Recovery Architectures

Figure 26 Extended Distance Cluster with Two Data Centers and DWDM

Node A

Network Switch Network Switch

Node B

Cluster

Heartbeat

Network and

Storage

Connections

Data

replication

is being

done with

MirrorDisk/UX

Maximum

distance

100 kilometers

Node C Node D

Storage Storage

Cluster Lock

Site 1 Site 2

FC Switch

Cluster Lock

FC Switch

DWDMDWDM

DWDM DWDM

Network Switch Network Switch

In the Figure 26 (page 64), the alternate paths for both networking and Fibre Channel are configured

through separate redundant pairs of DWDM boxes. Each Fibre Channel switch has one ISL

connected to each DWDM box. You can also use a single DWDM box in each data center.The

DWDM boxes must be internally redundant and have no SPOFs, and have at least two dark fiber

links between them, each routed geographically differently, and the redundant standby fiber feature

is configured for the DWDM boxes. This cluster is not allowed to contain more than four nodes

with Serviceguard. For EC RAC clusters, the number of nodes allowed and the maximum distance

supported vary, depending upon the HP UX revision and the volume manager being used (see

Table 3 (page 49), Table 4 (page 49) and Table 5 (page 51) for details). Note that you cannot

use SONET instead of DWDM in this configuration, as SONET is not supported with dual cluster

lock disks.

64 Extended Distance Cluster Configurations