How do node and zone work in Linux?
And why Linux use them?
I really do not anything about them.

The are used to identify what piece of physical memory a page is mapped to. Differant physical memory ranges has differant characteristics. For instance, some ranges permit DMA transfers while others do not.

Try google "linux zome node" for more info. here's a good link: http://home.earthlink.net/~jknapka/l...zonealloc.html

there are three zones:

ZONE_DMA First 16MiB of memory
ZONE_NORMAL 16MiB - 896MiB
ZONE_HIGHMEM 896 MiB - End

this is to specify how memory can be used and what memory is to be used for what...and of course there are fallovers into other territories when under pressure...

node can be looked at as a container that contains all the pages..for different zones...these nodes are represented by data structures: pg_data_t..there may be multiple such nodes...
i would suggest you go through this great book:
Understanding The Linux Virtual Memory Manager:Mel Gorman
there is an annotated code walkttrough for this book too:
Code Commentary On The Linux Virtual Memory Manager

these are exhaustive works..so you would find answers to all
your questions in these books

thank you