Compaq ProLiant 8500 Compaq Parallel Database Cluster Model PDC/05000 for Orac - Page 81
Planning Cluster Configurations for Redundant Fibre Channel Fabrics
View all Compaq ProLiant 8500 manuals
Add to My Manuals
Save this manual to your list of manuals |
Page 81 highlights
Cluster Planning 4-7 Planning Cluster Configurations for Redundant Fibre Channel Fabrics A redundant Fibre Channel Fabric is two or more Fibre Channel SAN Switches installed between host adapters in cluster nodes and array controllers in the shared storage subsystems. Each redundant Fibre Channel Fabric in a PDC/O5000 provides I/O path connections between each cluster node and the two ports on both array controllers in each MA8000/EMA12000 Storage Subsystem. Thus, each node can issue I/O requests to every storageset in the storage subsystems contained in that redundant Fibre Channel Fabric. A redundant Fibre Channel Fabric increases a PDC/O5000 cluster's availability by providing redundant components along the I/O paths between the cluster nodes and the storage subsystems. If a particular component fails, data communication can continue over an I/O path containing functional components. To achieve redundancy in Fibre Channel Fabrics, you can implement either a dual redundancy configuration or a quad redundancy configuration. Planning Dual Redundancy Configurations A dual redundancy configuration is the minimum allowable configuration that provides redundancy along the I/O paths between the cluster nodes and the storage subsystems. This configuration provides two of each component along the I/O paths. These components include: I At least two cluster nodes I Two host adapters in each node I Two Fibre Channel SAN Switches I Two dual-port array controllers in each storage subsystem I Fibre Channel cables and GBIC-SW modules for the connections between the host adapters and the Fibre Channel SAN Switches, and between the Fibre Channel SAN Switches and the array controllers If a redundant component on one of the I/O paths fails, the cluster nodes can continue to access shared storage through the path containing the surviving component.