This topic includes information about the switches that are supported on the SAN.
The SAN must contain only supported switches. The SAN Volume Controller supports specific IBM 2109, McData, and InRange switch models and the Cisco MDS 9000 switch and switches supported by the Cisco MDS 9000.
See the following Web site for specific firmware levels and the latest supported hardware: http://www.ibm.com/storage/support/2145/
Operation with other switches is not supported.
Different vendor switches cannot be intermixed in the same counterpart SAN. A redundant SAN, made up of more than one counterpart SAN can contain different vendor switches, provided the same vendor is used within each counterpart SAN.
The SAN must consist of two independent switches (or networks of switches) so that the SAN includes a redundant fabric, and has no single point of failure. If one SAN fabric fails, the configuration is in a degraded mode, but it is still valid. If the SAN contains only one fabric, it is still a valid configuration, but a failure of the fabric might cause a loss of access to data. Such a SAN, therefore, is seen as a single point of failure.
Configurations with more than two SANs are not supported.
On the fibre-channel SAN, the SAN Volume Controller nodes must always and only be connected to SAN switches. Each node must be connected to each of the counterpart SANs that are in the redundant fabric. Any operation that uses direct connections between host and node, or controller and node, is not supported.
On the fibre-channel SAN, back-end storage must always and only be connected to SAN switches. Multiple connections are permitted from the redundant controllers of the back-end storage, to improve data bandwidth performance. It is not necessary to have a connection between each redundant disk controller system of the back-end storage and each counterpart SAN. For example, in a FAStT configuration in which the FAStT contains two redundant controllers, only two controller minihubs are usually used. Controller A of the FAStT is, therefore, connected to counterpart SAN A, and controller B of the of the FAStT is connected to counterpart SAN B. Any operation that uses a direct connection between the host and the controller is not supported.
The connections between the switches and the SAN Volume Controllers can operate at 1 Gbps or at 2 Gbps. All the ports for the SAN Volume Controllers that are in a single cluster, however, must run at one speed. Any operation that runs different speeds on the node-to-switch connections that are in a single cluster is not valid.
| Attention: The default transfer rate in the SAN Volume Controller is 2 Gbps. If your environment is set up to use 1 Gbps switches, the switch rate must be set at the transfer rate. |
Mixed speeds are permitted in the fabric. Lower speeds can be used to extend distances or to make use of 1 Gbps legacy components.
The switch configuration of a SAN Volume Controller SAN must observe the switch manufacturer's configuration rules. These rules might put restrictions on the switch configuration; for example, the switch manufacturer might not permit other manufacturer's switches to be in the SAN. Any operations that run outside the manufacturer's rules is not supported.
The switch must be configured so that the SAN Volume Controller nodes can see the back-end storage and the front-end HBAs. However, the front-end HBAs and the back-end storage must not be in the same zone. Any operation that runs outside these zoning rules is not supported.
Because each SAN Volume Controller has four ports, the switches can be zoned so that a particular SAN Volume Controller port is used only for internode communication, for communication to the host, or for communication to back-end storage. Whatever the configuration, each SAN Volume Controller node must remain connected to the full SAN fabric. Zoning must not be used to split the SAN into two parts.
With Remote Copy, additional zones are required that contain only the local nodes and the remote nodes. It is valid for the local hosts to see the remote nodes, or for the remote hosts to see the local nodes. Any zone that contains the local and the remote back-end storage and local nodes or remote nodes, or both, is not valid.
The local or remote fabric must not contain more than three Inter-Switch Links (ISLs) in each fabric. Any operation that uses more than three ISLs is not supported. When a local fabric is connected to a remote fabric for Remote Copy purposes, the ISL count between a local node and a remote node must not exceed seven. Therefore, some ISLs can be used in a cascaded switch link between local and remote clusters if the internal ISL count of the local or remote cluster is less than three.
The local and remote fabric interconnections must be only one ISL hop between a switch that is in the local fabric and a switch that is in a remote fabric. That is, it must be a single-mode fibre up to 10 KM (32 810 ft.) long. Any operation that uses other local or remote fabric interconnections is not supported.
When ISLs are used, each ISL oversubscription must not exceed six. Any operation that uses higher values is not supported.
With Inter-Switch Links between nodes in the same cluster, the ISLs are considered a single point of failure. This is illustrated in Figure 37.

If Link 1 or Link 2 fails, the cluster communication does not fail.
If Link 3 or Link 4 fails, the cluster communication does not fail.
If ISL 1 or ISL 2 fails, the communication between Node A and Node B will fail for a period of time, and the node will not be recognized, even though there is still a connection between the nodes.
To ensure that a fibre-channel link failure does not cause nodes to fail when there are ISLs between nodes, it is necessary to use a redundant configuration. This is illustrated in Figure 38.

With a redundant configuration, if any one of the links fails, then communication on the cluster will not fail.