[gpfsug-discuss] sequential I/O write - performance

[Jonathan Buzzard]

On 10/02/2024 19:39, dale mac wrote:

[SNIP]

> *What to do,*
> 
> **
> 
> *Try *
> 
> 
>  1. 8 individual gpfs FS with your FIO test like XFS test. Ie do like
>     for like. *8x XFS versus 8x GPFS*.  From an array perspective same
>     io pattern.
>  2. 1 gpfs FS and 1 Array and matching 1 FIO ( then x8 result)
> 

Pull in the eight devices, using LVM create PV's on them, and then 
combine the eight PV's into a single VG, make a single LV taking up all 
the VG, and then create a single XFS file system on the LV.

You want to use "-i 8" when creating the LV and experiment with the -I 
option to set the size of the stripe.

You now have an XFS file system that matches the GPFS file system. Rerun 
your tests comparing apples to apples rather than oranges.


JAB.

-- 
Jonathan A. Buzzard                         Tel: +44141-5483420
HPC System Administrator, ARCHIE-WeSt.
University of Strathclyde, John Anderson Building, Glasgow. G4 0NG