Brocade Access Gateway Administrator's Guide Supporting Fabric OS v7.0.0 (53-1002156-01, April 2011)

Access Gateway Administrator’s Guide 23
53-1002156-01
Access Gateway mapping
2
Pre-provisioning
You can use Fabric OS commands, Web Tools, and Fabric Manager to map devices that do not yet
exist. This allows applicable management programs to push configuration changes without
worrying about the order in which they are received. For example, if system administrators need to
push a set of port group changes and a set of device mapping changes, they could push them in
either order without error. This also applies to using Fabric OS commands for device mapping. You
could also map several devices to a new port group and then create the group without error. You
can also remove one device, and then remove another device without error.
VMware configuration considerations
Enabling device mapping for individual virtual machines (VMs) running on a VMware ESX server
connected to an F_Port can redirect I/O traffic for these VMs, provided the server is configured to
use Raw Device Mapped storage. All traffic will originate from a VM’s WWN and will follow any
mapping configured for the WWN. If anything interrupts the virtual port’s connection for the VM,
such as a failover port being used because a port goes offline, traffic will originate from the ESX
server’s base device port ID and not the VM’s port ID. If there are any additional disruptions, the
server will not switch back to the virtual port, and the VM’s traffic will not follow the configured
device mapping. Note that this can also occur when a VM first boots, prior to any failover.
When this behavior occurs, the VM’s WWN will be properly logged in to the fabric. The WWN
appears in the output of ag --show and ag --wwnmapshow, as well as on the switch. The output from
the portperfshow command displays all traffic on the port to which the ESX server port is mapped
(base PID).
Configuring device mapping
To configure WWN mapping on VMware ESX systems, use the following steps.
1. Make sure that virtual world wide port names (VWWPN) of virtual machines (VMs) are mapped
to the correct port group (or N_Port). Map all VWWPNs to N_Ports to avoid confusion.
2. Make sure all VWWPNs are mapped for LUN access for array-based targets.
3. Make sure to include all VWWPNs in the zone configuration.
4. Reboot the VM.
5. Zone the server’s physical port to the storage device.
6. Check the traffic that originates from the virtual node PID (VN PID). If the configuration is
correct, traffic will flow from the VN PID.
For additional information on using device mapping for connecting VMware systems, refer to the
Technical Brief How to Configure NPIV on VMware ESX Server 3.5 at following link:
http://www.brocade.com/downloads/documents/brocade_vmware_technical_briefs/Brocade_NP
IV_ESX3.5_WP.pdf.
Failover and failback considerations
When using device mapping with VMware, the base device initiates PLOGI and PRLI to the target,
and then discovers the LUN. The virtual device also initiates a PLOGI and PRLI to the target, but
LUN discovery does not occur. Therefore, when the device-mapped port is toggled and failover or
failback takes place, traffic will resume from the base device. One of the following actions is
recommended when using device mapping with VMware: