:
LVM Working Theorem
When
LVM prepares to store data into one LV,
how does it determine which real disk(s) it should choose? In LVM, one PV
comprises several PEs (Physical Extents) and one LV comprises several LEs (Logical Extents).
These PEs map LEs directly and their translating relation is stored into a 'PE/LE
Translation Table' (situated in VGDA). Stored in LVM disk, Translation Table
will be loaded to internal memory as soon as VG is activated. PE is appointed
when creating VG with 4M
default size; Whether or not the size and type of the real hard disk are the
same, all PEs are as big as each other in one VG. LV
being created, LVM system creates LE with automatically maintenance of PE/LE
Translation Table so that each LE in LV
can find its mapping PE. In conclusion, LVM system knows where to write down
disk data.