The Dell EqualLogic PS-Series arrays physically connect to your cluster using ethernet connections. Each controller in a PS-Series array has three ports, which are used for data and management connections. All of the ports from both controllers should all be attached to the same subnet.
Cabling the Power Supplies
To ensure that the specific power requirements are satisfied, see the documentation for each component in your cluster solution. It is recommended that you adhere to the following guidelines to protect your cluster solution from power-related failures:
- Plug redundant power supplies from each cluster node or storage array into separate AC circuits.
- Plug cluster node power supplies into separate optional network power switches.
- Use uninterruptible power supplies (UPS).
- Consider backup generators and power from separate electrical substations.
Cabling the iSCSI Ports
To cable the array:
All of the iSCSI interfaces on the PS-Series array must be configured on the same network segment. To provide redundant connection paths and fault resilience in the event of a switch failure, a two-switch iSCSI network topology is recommended. In such a configuration, there should be multiple ports with link aggregation providing the inter-switch, or trunk, connection. Further, each EqualLogic controller should have two connections to one switch and one connection to the other switch. The active and stand-by controller in the same member should have dual connections to alternate switches. This topology is illustrated for a single PS5000 array in the figure below. File:EqualLogic two switches.png
At least two network adapter ports on each host server should be configured for iSCSI, with one of these ports connected to each of the two switches. If any component in the storage path -- such as the iSCSI NIC, the cable, the switch, or the storage controller -- fails, multipath software automatically re-routes the I/O requests to the alternate path so that the storage array and cluster node continue to communicate without interruption. A cluster node with 2 single-port NICs can provide higher availability; a single NIC failure in such a configuration does not cause cluster resources to move to another cluster node.
Continue to Management Node Configuration