Re: raid5 software vs hardware: parity calculations?

[Bill Davidsen <davidsen@tmr.com>]

Robin Bowes wrote:
> Bill Davidsen wrote:
>   
>> There have been several recent threads on the list regarding software
>> RAID-5 performance. The reference might be updated to reflect the poor
>> write performance of RAID-5 until/unless significant tuning is done.
>> Read that as tuning obscure parameters and throwing a lot of memory into
>> stripe cache. The reasons for hardware RAID should include "performance
>> of RAID-5 writes is usually much better than software RAID-5 with
>> default tuning.
>>     
>
> Could you point me at a source of documentation describing how to
> perform such tuning?
>   
No. There has been a lot of discussion of this topic on this list, and a 
trip through the archives of the last 60 days or so will let you pull 
out a number of tuning tips which allow very good performance. My 
concern was writing large blocks of data, 1MB per write, to RAID-5, and 
didn't involve the overhead of small blocks at all, that leads through 
other code and behavior.

I suppose while it's fresh in my mind I should write a script to rerun 
the whole write test suite and generate some graphs, lists of 
parameters, etc. If you are writing a LOT of data, you may find that 
tuning the dirty_* parameters will result in better system response, 
perhaps at the cost of some small total write throughput, although I 
didn't notice anything significant when I tried them.
> Specifically, I have 8x500GB WD STAT drives on a Supermicro PCI-X 8-port
> SATA card configured as a single RAID6 array (~3TB available space)
>   
No hot spare(s)?

-- 
bill davidsen <davidsen@tmr.com>
  CTO TMR Associates, Inc
  Doing interesting things with small computers since 1979