Gryn wrote:
>
hi,
> Ew,
> There is a slight preformance drop. This comes from swap-files having to
> access an additional layer -- the file system. Swap partitions can be read
> directly from the harddrive. Swapfiles could also become fragmented, but
> this generally isn't much of an issue either.
is there a possibility to tell the filesystem to store it a continous
block
(which is perhaps aligned to the underlying device's block size)
a kind of virtual partition within a filesystem.
i imagine it like this:
a filesystem may have an additional interface/ioctl for raw access
(set file to raw-access, get device, get blocksize, get/set blockcount,
get fragmentation info, ...)
so an application can get definitive information where the file's data
are exactly stored.
now the loopdev can use this interface to improve the access
performance.
> The way I feel about it is that if you are using larges amount of swap, then
> you are doing too much with your system. If you have seen any chart plotting
> "work done" versus "resources used", you see if you use swap space too
> heavily, then preformance drops off almost to zero!!! (This is called
> thrashing).
yeah, thats the risk of huge swapping. but without this, the apps would
simple
crash - so you have the control which application you kill (or if you
better go drinking a cup of coffee :) )
bye,
ew.
---------------------------------------------
no mercy for M$
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Fri Jun 23 2000 - 21:00:25 EST