Follow up : Setup new System on PI1LAP

This is a follow up on this post

http://packet-radio.net/2017/12/setup-new-system-on-pi1lap/

Today the SSD’s are installed in the PC. I ran into quite a few problems. Hardware-based raid is not possible with the PCI card and SSDs. Now the mainboard can also raid. Unfortunately, the SSDs are not recognized here. Cheap is expensive. So now over to software raid. Software raid runs perfectly. But here I have to hand in speed.

Here below the benchmark of the old sata hard disks. The old sata 80Gb disks were in Hardware-based raid. Cheap pci card from Delock. Both systems use raid 1.

I use the same command for both tests.

./fio --randrepeat=1 --ioengine=libaio --direct=1 --gtod_reduce=1 --name=test --filename=test / 
--bs=4k --iodepth=64 --size=4G --readwrite=randrw --rwmixread=75
Starting 1 process
Jobs: 1 (f=1): [m] [100.0% done] [1504KB/388KB/0KB /s] [376/97/0 iops] [eta 00m:00s]
test: (groupid=0, jobs=1): err= 0: pid=8223: Thu Dec 7 08:54:51 2017
read : io=3071.7MB, bw=498349B/s, iops=121, runt=6463092msec
write: io=1024.4MB, bw=166188B/s, iops=40, runt=6463092msec
cpu : usr=0.11%, sys=0.58%, ctx=1035400, majf=0, minf=6
IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%
submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.1%, >=64=0.0%
issued : total=r=786347/w=262229/d=0, short=r=0/w=0/d=0
latency : target=0, window=0, percentile=100.00%, depth=64
 
Run status group 0 (all jobs):
READ: io=3071.7MB, aggrb=486KB/s, minb=486KB/s, maxb=486KB/s, mint=6463092msec, maxt=6463092msec
WRITE: io=1024.4MB, aggrb=162KB/s, minb=162KB/s, maxb=162KB/s, mint=6463092msec, maxt=6463092msec
 
Disk stats (read/write):
sda: ios=782814/264536, merge=3607/1804, ticks=309674428/102480492, in_queue=412156584, util=100.00%

Here you see a read of 121 iops (Input/Output Operations per Second) and a write of 40 iops.

Now here is a benchmark with the ssd’s.

Starting 1 process
test: Laying out IO file(s) (1 file(s) / 4096MB)
Jobs: 1 (f=1): [m] [100.0% done] [19988KB/6481KB/0KB /s] [4997/1620/0 iops] [eta 00m:00s]
test: (groupid=0, jobs=1): err= 0: pid=29862: Sat Dec  9 04:51:12 2017
  read : io=3071.7MB, bw=15291KB/s, iops=3822, runt=205708msec
  write: io=1024.4MB, bw=5099.6KB/s, iops=1274, runt=205708msec
  cpu          : usr=2.56%, sys=14.12%, ctx=214110, majf=0, minf=5
  IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.1%, >=64=0.0%
     issued    : total=r=786347/w=262229/d=0, short=r=0/w=0/d=0
     latency   : target=0, window=0, percentile=100.00%, depth=64

Run status group 0 (all jobs):
   READ: io=3071.7MB, aggrb=15290KB/s, minb=15290KB/s, maxb=15290KB/s, mint=205708msec, maxt=205708msec
  WRITE: io=1024.4MB, aggrb=5099KB/s, minb=5099KB/s, maxb=5099KB/s, mint=205708msec, maxt=205708msec

Disk stats (read/write):
  md0: ios=786158/262297, merge=0/0, ticks=0/0, in_queue=0, util=0.00%, aggrios=392956/262257, aggrmerge=217/165, aggrticks=4046074/3411330, aggrin_queue=7456728, aggrutil=91.99%
  sda: ios=393413/262260, merge=238/162, ticks=4029372/3405552, in_queue=7434076, util=91.15%
  sdb: ios=392499/262254, merge=197/168, ticks=4062776/3417108, in_queue=7479380, util=91.99%

Here you can see that the iops are much more. So we make progress. It`s not much iops of 3800 for a ssd. But probably is the software raid the cause. And the old system perhaps.

md0 is the array and the sda and sdb are the ssd disk in the array.

I have read several tests about SSD hard disks. If I write 10Gb to the SSD per day, I can probably do 209 years with it. Let’s take 5% of that, is still more then 10 years. But yes you never know with a hard disk.

mdadm is running in the background to monitor the array. Is there something wrong I get a email.

This is a nice website about software raid.
https://raid.wiki.kernel.org/index.php/RAID_setup

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.