In a world where everything is 100% perfectly configured, you might be able to unmap the LUN, remap the LUN with a new ID, rescan for disks and datastores at all the ESX hosts, and get by with your guests not noticing.
DON'T TRY IT.
Something in your environment isn't setup get through that procedure, and by definition it will cause a problem with what you can least afford to lose.
Pick a LUN - shutdown the guests that use the LUN, unmap, remap, rescan for disks and datastores, restart the affected guests.
Alternatively, vMotion all the guests that use the LUN onto the host or hosts that has the same LUN with the ID you want. Then do the unmap/remap/rescan bit on just the one host that isn't using the LUN. Then you can rebalance back. But be sure you know what you are doing - you need to turn off anything that might automatically try to migrate load back to the one host while you are messing with it.
I prefer the shutdown mechanism for this, since you can then fix up all ESX hosts that might need changing in one go.
