On Wed, Mar 22, 2006 at 11:37:22PM +0100, Simon Gabor Technical Support
Engineer wrote:
> RAID-1-nel az olvasasi sebesesseg elvileg duplaja lehetne a fizikai
> diszk-nel (ha ket fizikai diszk alkotja a RAID-1-et), hiszen a 2 fizikai
> vincsit parhuzamosan (round-robin) lehetne olvasni az adatokert.
> (Elso szektort az elso vincsi, a masodikat a masodik vincsi olvassa ki,
> idoben _egyszerre_.)
Egyetlen aprocska iciri-piciri problema, hogy a szektor atugrasahoz
szukseges seek _sokkal_ tovabb tart, mint egyszeruen ugyanarrol a
diszkrol folyamatosan olvasni a kovetkezo szektort. Emiatt szekvencialis
olvasast a RAID1 nem tud gyorsitani (valamikor regen a linux-kernelen
volt errol egy thread; ha bovebben erdekel, turkalj az archivumban).
A random I/O-t lehet gyorsitani, es valamennyire mukodik is, lasd kernel
forrasban drivers/md/raid1.c:
/*
* This routine returns the disk from which the requested read should
* be done. There is a per-array 'next expected sequential IO' sector
* number - if this matches on the next IO then we use the last disk.
* There is also a per-disk 'last know head position' sector that is
* maintained from IRQ contexts, both the normal and the resync IO
* completion handlers update this position correctly. If there is no
* perfect sequential match then we pick the disk whose head is closest.
*
* If there are 2 mirrors in the same 2 devices, performance degrades
* because position is mirror, not device based.
*
* The rdev for the device selected will have nr_pending incremented.
*/
static int read_balance(conf_t *conf, r1bio_t *r1_bio)
{
[...]
Gabor
--
---------------------------------------------------------
MTA SZTAKI Computer and Automation Research Institute
Hungarian Academy of Sciences
---------------------------------------------------------
_________________________________________________
linux lista - [email protected]
http://mlf2.linux.rulez.org/mailman/listinfo/linux
