certance DAT

Glenn Dawson glenn at antimatter.net
Wed Jul 27 04:23:39 GMT 2005


At 08:17 PM 7/26/2005, Chuck Swiger wrote:
...so it's not exactly super-zippy, either.  Hmm, is it just me, or are the 
following numbers significantly low for a RAID-1 of two 10K RPM U320 SCSI 
disks...?

>/dev/amrd1
>         512             # sectorsize
>         73274490880     # mediasize in bytes (68G)
>         143114240       # mediasize in sectors
>         8908            # Cylinders according to firmware.
>         255             # Heads according to firmware.
>         63              # Sectors according to firmware.
>
>Seek times:
>         Full stroke:      250 iter in   1.824059 sec =    7.296 msec
>         Half stroke:      250 iter in   1.805398 sec =    7.222 msec
>         Quarter stroke:   500 iter in   4.254147 sec =    8.508 msec
>         Short forward:    400 iter in   2.821081 sec =    7.053 msec
>         Short backward:   400 iter in   2.860203 sec =    7.151 msec
>         Seq outer:       2048 iter in   8.821875 sec =    4.308 msec
>         Seq inner:       2048 iter in   9.006505 sec =    4.398 msec
>Transfer rates:
>         outside:       102400 kbytes in   9.242111 sec =    11080 kbytes/sec
>         middle:        102400 kbytes in   9.230325 sec =    11094 kbytes/sec
>         inside:        102400 kbytes in  10.779231 sec =     9500 kbytes/sec
>
>[ This is running RELENG_5_4... ]

I would have expected the transfer rates to be about twice what they're 
listed as here.  Though I don't know how you measured them.

If you want to see something interesting, create a ufs1 file system on the 
same raid 1 using FreeBSD 4.x.  Then mount it in 5.x and do your test for 
transfer rates.  Compare that to either ufs1 or ufs2 on the same raid 1 as 
created by 5.4.

The results are very interesting.

-Glenn


>--
>-Chuck



More information about the freebsd-questions mailing list