[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: MySQL Performance Woody Vs Sarge

On Wed, Jun 01, 2005 at 06:03:28PM +1200, Simon wrote:
> Jacob S wrote:
> 
> The above raid consists of:
> 
> # hdparm /dev/hde
> 
> /dev/hde:
>  multcount    =  0 (off)
>  I/O support  =  0 (default 16-bit)
>  unmaskirq    =  0 (off)
>  using_dma    =  1 (on)
>  keepsettings =  0 (off)
>  nowerr       =  0 (off)
>  readonly     =  0 (off)
>  readahead    =  8 (on)
>  geometry     = 5005/255/63, sectors = 80418240, start = 0
>  busstate     =  1 (on)
> 
> # hdparm /dev/hdg
> 
> /dev/hdg:
>  multcount    =  0 (off)
>  I/O support  =  0 (default 16-bit)
>  unmaskirq    =  0 (off)
>  using_dma    =  1 (on)
>  keepsettings =  0 (off)
>  nowerr       =  0 (off)
>  readonly     =  0 (off)
>  readahead    =  8 (on)
>  geometry     = 79780/16/63, sectors = 80418240, start = 0
>  busstate     =  1 (on)
> 
> 
> sarge(production) = slower (Linux version 2.6.8-2-k7):
> 
> # hdparm /dev/md0
> /dev/md0:
>  readonly     =  0 (off)
>  readahead    = 256 (on)
>  geometry     = 25248/2/4, sectors = 80418048, start = 0
> 
> The above raid consists of:
> 
> # hdparm /dev/hde
> 
> /dev/hde:
>  multcount    =  0 (off)
>  IO_support   =  0 (default 16-bit)
>  unmaskirq    =  0 (off)
>  using_dma    =  1 (on)
>  keepsettings =  0 (off)
>  readonly     =  0 (off)
>  readahead    = 256 (on)
>  geometry     = 65535/16/63, sectors = 80418240, start = 0
> 
> # hdparm /dev/hdg
> 
> /dev/hdg:
>  multcount    =  0 (off)
>  IO_support   =  0 (default 16-bit)
>  unmaskirq    =  0 (off)
>  using_dma    =  1 (on)
>  keepsettings =  0 (off)
>  readonly     =  0 (off)
>  readahead    = 256 (on)
>  geometry     = 65535/16/63, sectors = 80418240, start = 0
> 

Hello,

it looks like an old issue of readahead. The units have changed in the
2.6 kernel, so you might experiment with several values to get the same
performance as in 2.4. My drive reports

readahead    = 8192 (on)

for the moment.

For what I can remember from the linux-kernel mail list, the argument
to readahead made a big difference on the performance of the drives.

Just my 0.02€

Ionut

-- 
***************
* Ionuţ Georgescu
* Max-Planck-Institut für Physik komplexer Systeme
* Noethnitzer Str. 38, D-01187 Dresden
* Phone: +49 (351) 871-2209
* Fax:   +49 (351) 871-1999

Attachment: signature.asc
Description: Digital signature

Reply to: