tech-kern archive

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

Re: WAPL/RAIDframe performance problems

> Is it possible for you to set up a test with raid5 using a different sata 
> or sas interface card?
I tried a Level 1 RAID on two SATA discs (on the same hardware that I did the 
plain-SATA test on).

Throughput to the fs is 83MB/s write, 101MB/s read.
svn update takes 5,6s without WAPBL and 7s with WAPBL.
The touch..rm test takes 15,1s without WAPL and 13,6s with WAPBL.
Compared to plain disc on the same hardware:
- FS throughput is roughly as on a plain disc, as expected.
- svn update is slowed down only with WAPBL enabled.
- touch..rm performs roughly equivalent (to plain disc) without WAPBL,
  but takes three times as long as on a plain disc with WAPBL enabled.
Compared to the mpt(4) SAS Level 1 RAID:
- FS write throughput is much faster as on mpt
- svn update with WAPBL is twice as fast as on mpt (no difference to mpt
  when WAPBL is disabled)

I still don't understand why writing to a RAID 1 in stripe-sized chunks
could be noticebly slower than writing to the component device (as it is on
my mpt(4) SAS drives).

Home | Main Index | Thread Index | Old Index