Re: Object Write Latency

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

 



Some more tests...

~journal in XFS:
[root@p05151113777233 fio]# time dd if=/dev/zero
of=/var/lib/ceph/osd/osd.1045/testtest bs=512 count=1 conv=fsync
1+0 records in
1+0 records out
512 bytes (512 B) copied, 0.0423365 s, 12.1 kB/s

real 0m0.044s
user 0m0.000s
sys 0m0.001s


~directio via the FS
[root@p05151113777233 fio]# time dd if=/dev/zero
of=/var/lib/ceph/osd/osd.1045/testtest bs=512 count=1
oflag=direct,dsync
1+0 records in
1+0 records out
512 bytes (512 B) copied, 0.0426422 s, 12.0 kB/s

real 0m0.044s
user 0m0.000s
sys 0m0.001s


~directio to the bdev seems to be faster (a few attempts below):

[root@p05151113760120 ~]# time dd if=/dev/zero of=/dev/sdo1 bs=512
count=1 seek=$(($RANDOM*1000)) oflag=direct,dsync
1+0 records in
1+0 records out
512 bytes (512 B) copied, 0.0113362 s, 45.2 kB/s

real 0m0.013s
user 0m0.001s
sys 0m0.001s
[root@p05151113760120 ~]# time dd if=/dev/zero of=/dev/sdo1 bs=512
count=1 seek=$(($RANDOM*1000)) oflag=direct,dsync
1+0 records in
1+0 records out
512 bytes (512 B) copied, 0.0129222 s, 39.6 kB/s

real 0m0.014s
user 0m0.001s
sys 0m0.000s
[root@p05151113760120 ~]# time dd if=/dev/zero of=/dev/sdo1 bs=512
count=1 seek=$(($RANDOM*1000)) oflag=direct,dsync
1+0 records in
1+0 records out
512 bytes (512 B) copied, 0.00484823 s, 106 kB/s

real 0m0.006s
user 0m0.001s
sys 0m0.001s
[root@p05151113760120 ~]#
[root@p05151113760120 ~]# time dd if=/dev/zero of=/dev/sdo1 bs=512
count=1 seek=$(($RANDOM*1000)) oflag=direct,dsync
1+0 records in
1+0 records out
512 bytes (512 B) copied, 0.00441663 s, 116 kB/s

real 0m0.006s
user 0m0.001s
sys 0m0.001s


So it looks like I really should try those O_DIRECT journals on a
separate partition.

Cheers, Dan

On Tue, Sep 24, 2013 at 11:13 AM, Dan van der Ster <dan@xxxxxxxxxxxxxx> wrote:
> On Mon, Sep 23, 2013 at 8:18 PM, Sage Weil <sage@xxxxxxxxxxx> wrote:
>> You
>> might try measuring that directly and comparing it to the 33ms
>> append+fsync that you previously saw.
>
> dd with fsync is quite slow...
>
> [root@p05151113777233 fio]# time dd if=/dev/zero
> of=/var/lib/ceph/osd/osd.1045/testtest bs=5 count=1
> 1+0 records in
> 1+0 records out
> 5 bytes (5 B) copied, 8.884e-05 s, 56.3 kB/s
>
> real 0m0.001s
> user 0m0.000s
> sys 0m0.001s
>
>
> [root@p05151113777233 fio]# time dd if=/dev/zero
> of=/var/lib/ceph/osd/osd.1045/testtest bs=5 count=1 conv=fsync
> 1+0 records in
> 1+0 records out
> 5 bytes (5 B) copied, 0.0435103 s, 0.1 kB/s
>
> real 0m0.045s
> user 0m0.001s
> sys 0m0.001s
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html




[Index of Archives]     [CEPH Users]     [Ceph Large]     [Information on CEPH]     [Linux BTRFS]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]
  Powered by Linux