Strange IO performance with UFS
Stefan Parvu
sparvu at systemdatarecorder.org
Fri Jul 4 21:19:57 UTC 2014
Hi,
> > I'm doing some tests on IO performance using fio, and I've found
> > something weird when using UFS and large files. I have the following
> > very simple sequential fio workload:
System:
FreeBSD ox 10.0-RELEASE-p6 FreeBSD 10.0-RELEASE-p6 #0: Tue Jun 24 07:47:37 UTC 2014
root at amd64-builder.daemonology.net:/usr/obj/usr/src/sys/GENERIC amd64
1. Seq Write to 1 file, 10GB size, single writer, block 4k, UFS2:
I tried to write seq using a single writer using an IOSIZE similar to your example, 10
GB to a 14TB Hdw RAID 10 LSI device using fio 2.1.9 under FreeBSD 10.0.
Result:
Run status group 0 (all jobs):
WRITE: io=10240MB, aggrb=460993KB/s, minb=460993KB/s, maxb=460993KB/s,
mint=22746msec, maxt=22746msec
2. Seq Write to 2500 files, each file 5MB size, multiple writers, UFS2:
Result:
Run status group 0 (all jobs):
WRITE: io=12500MB, aggrb=167429KB/s, minb=334KB/s, maxb=9968KB/s,
mint=2568msec, maxt=76450msec
Questions:
- where are you writing, what storage: hdw / sfw RAID ?
- are you using time based fio tests ?
For fun I can share with you some results we been doing between FreeBSD10 amd64 (f10)
and Debian7 amd64 (d7) using LSI HDW RAID 10. We don't use time based fio but rather
we measure how fast we can send once the IOSIZE and measure the elapsed time.
This proofed to be more accurate and return more sane results than actually keeping
fio running for 15 or 30minutes.
Id Test_Name Throughput Utilization Idle
1 f10.raid10.4k.2500 23 MB/s 8% 92%
2 f10.raid10.4k.5000 18 MB/s 9% 91%
3 f10.raid10.64k.2500 215 MB/s 22% 78%
4 f10.raid10.64k.5000 162 MB/s 18% 82%
idle iowait
5 d7.raid10.4k.2500 29 MB/s 2% 65.08 + 32.93
6 d7.raid10.4k.5000 29 MB/s 3% 53.68 + 43.79
7 d7.raid10.64k.2500 297 MB/s 3% 56.44 + 41.11
8 d7.raid10.64k.5000 182 MB/s 4% 12.85 + 83.85
--
Stefan Parvu <sparvu at systemdatarecorder.org>
More information about the freebsd-fs
mailing list