Hi (again) Brian,
A LUN (Logical Unit Number) & is a subdivision of a larger storage device. Typically a RAID array will be "sliced" up into multiple LUNs *almost* always smaller than disk capacities. Sometimes a LUN can be mult drives chained together to a capacity larger than the disks themselves. But the point is that the OS sees a LUN as a distinct disk. It knows not just how it's formed...it thinks it's a disk.
Now let's say that the app exists now at /opt/app. What you can do is backup that dir & the tree on down, delete the files in it & all subdirs. Create the new VG/LV/FS & mount this new structure right to /opt/app. Then restore the data back to it & the app should be none the wiser.
What JP states, while fundamentally correct, is not the way I would do it. I want to have separate VGs & I want MC/SG to mount/unmount them as the pkg comes up/goes down. I also like to have a virtual IP whether I *need* it or not. It's an easy way to check, from anywhere on the network & w/o logging into the server, if the pkg is up or down. Just ping the virtual IP & if it responds you know the pkg has not come down. Now...whether the app is running (properly) or not is entirely up to the monitoring mechanism employed. If you don't monitor it properly, the app could be in severe distress & the pkg may not go down & the virtual IP would still be available. Proper execution of monitoring is key to HA.
My original point was if these are distinct & separate pkgs & are not tightly integrated to the point where pkg B can't run w/o pkg A being up, then sep pkgs is the way to go.
Rgds,
Jeff
PERSEVERANCE -- Remember, whatever does not kill you only makes you stronger!