If you do, what you might well have is hdisks being collected together to form a vpath device. It is this device that you assign to your volume groups.
To expand on what rbatte1 said: when connecting virtual disks (LUNs) from a SAN to a system one usually has a "fabric" in between. Such a fabric is basically a FC-switch, redundant adapters/FC-lines to connect to the storage system (DS8000, EMC VMax, ....) and the clients (basically your systems).
Simple "hdisk"-devices are usually representing one such path over which a disk can be seen, but there might be several parallel ways to see the disk:
Here a single disk from the storage can be seen over line 1 then 3, 1 then 4, 2 then 3 or 2 then 4, so you'd have 4 hdisk devices which are just different views onto the same object. (Consider 4 pictures showing the same building but from different angles.) Multipath-drivers (MPIO, SDD, EMCs Powerpath, etc.) now bind together all these views to a usable pseudo-device representing the real disk behind all these. Depending on the driver they are named differently - vpath devices, powerpath devices, but the concept is the same.
Let's hope that the fibre ports in use at each end are not on the same card. Someone has done that here and which we're protected against the fibre cable failing, a failure in the port cannot be fixed without downtime, and a failure of the card causes downtime.
I have one card with dual FC port. What i'm trying to do is 2 paths trough each controller to access the same LUN. Then if one of the controller crash, i can still access the LUN with the other one.
Fibre channel storage uses a system of World Wide Names (WWN) to identify disks or other pieces of array. Some FC storage controllers allow you to assign (pseudo) WWN's to arrays created by the controller. Only one WWN should be allocated to any one piece of storage.
A modern host O/S storage driver will realize that when it can see the same WWN via different routes that it is actually the same storage that it can see through different routes and then manage that topology (by only using one route at a time unless the storage controller can accept MPIO, but also use an alternative route if the primary route fails).
You can't. You will have to install some sort of MPIO driver (which one that is depends on the SAN you use) and this driver will bundle several hdisk-devices (they represent the OSes view of the disk) to one multipath-device (vpath, powerpath, ..., these represent the multipath drivers view of the device)
See above, my rather exhaustive explanation of how a SAN works (#3): no! A WWN is like an address, not a way to get there. Regardless of the route you take it always ends at the same LUN. The WWN is an attribute of the LUN, not the path (otherwise all LUNS sharing the same path would have the same WWNs, no?)
AIX 5.3 should support this with the correct driver.
Have a look at this and this as a starting point on the IBM website. Does that give you an option?
I still worry that if you have a card failure, you lose your disks, or if you have a port failure, it requires downtime to replace. There is likely some performance overhead too, but it depends how hard the IO is being driven.
AIX only introduced support for the DS3400, with either native AIX MPIO or the (now obsolete) RDAC driver in AIX 5.3 TL06 so I'm afraid you are out of luck unless you update something.
Please also note that the DS3400 is only capable of active / passive control so all the multipath chat above doesn't really apply to your storage subsystem, each adapter should only be able to see one storage controller and as you have a direct connection - one fcs or fscsi device can only see one controller - your connections are fine.
