Terrible VM disk performance compared to LXC

M

murda

Member
Jan 3, 2024
51
4
8
I understand that LXC has less overhead than a VM, but the difference is so big that I suspect misconfiguration on the VM. Hopefully I can get help to correctly configure a VM for better performance.

LXC specs:
1722092350511.png

VM specs:
1722092388210.png

1722092414344.png

Host specs:
1722092491985.png
1722097559925.png

Using this fio configuration:
Code: 
[global]
ioengine=libaio
direct=1
time_based
runtime=180
group_reporting
filename=/tmp/test_file
size=10G

[random-read]
description=Simulates reading pieces from different parts of files
rw=randread
bs=256K
iodepth=32
numjobs=4
percentage_random=100

[random-write]
description=Simulates writing newly downloaded pieces
rw=randwrite
bs=256K
iodepth=16
numjobs=2
percentage_random=100

[mixed-readwrite]
description=Simulates simultaneous reading and writing
rw=randrw
bs=128K
rwmixread=70
iodepth=24
numjobs=3
percentage_random=100

[sequential-read]
description=Simulates reading larger sequential chunks
rw=read
bs=1M
iodepth=8
numjobs=1
percentage_random=0

[sequential-write]
description=Simulates writing larger sequential chunks
rw=write
bs=1M
iodepth=8
numjobs=1
percentage_random=0

LXC result:
Code: 
random-read: (g=0): rw=randread, bs=(R) 256KiB-256KiB, (W) 256KiB-256KiB, (T) 256KiB-256KiB, ioengine=libaio, iodepth=32
...
random-write: (g=0): rw=randwrite, bs=(R) 256KiB-256KiB, (W) 256KiB-256KiB, (T) 256KiB-256KiB, ioengine=libaio, iodepth=16
...
mixed-readwrite: (g=0): rw=randrw, bs=(R) 128KiB-128KiB, (W) 128KiB-128KiB, (T) 128KiB-128KiB, ioengine=libaio, iodepth=24
...
sequential-read: (g=0): rw=read, bs=(R) 1024KiB-1024KiB, (W) 1024KiB-1024KiB, (T) 1024KiB-1024KiB, ioengine=libaio, iodepth=8
sequential-write: (g=0): rw=write, bs=(R) 1024KiB-1024KiB, (W) 1024KiB-1024KiB, (T) 1024KiB-1024KiB, ioengine=libaio, iodepth=8
fio-3.37
Starting 11 processes
random-read: Laying out IO file (1 file / 10240MiB)

random-read: (groupid=0, jobs=11): err= 0: pid=697: Sat Jul 27 18:23:09 2024
  Description  : [Simulates reading pieces from different parts of files]
  read: IOPS=7374, BW=2038MiB/s (2138MB/s)(358GiB/180010msec)
    slat (usec): min=4, max=916643, avg=885.67, stdev=4949.76
    clat (msec): min=4, max=1040, avg=24.35, stdev=31.35
     lat (msec): min=4, max=1040, avg=25.24, stdev=32.11
    clat percentiles (msec):
     |  1.00th=[    9],  5.00th=[   10], 10.00th=[   13], 20.00th=[   16],
     | 30.00th=[   18], 40.00th=[   20], 50.00th=[   22], 60.00th=[   23],
     | 70.00th=[   25], 80.00th=[   27], 90.00th=[   29], 95.00th=[   33],
     | 99.00th=[  124], 99.50th=[  201], 99.90th=[  405], 99.95th=[  760],
     | 99.99th=[  944]
   bw (  MiB/s): min=   59, max= 2644, per=100.00%, avg=2052.03, stdev=106.38, samples=2856
   iops        : min=  224, max= 9688, avg=7423.90, stdev=324.22, samples=2856
  write: IOPS=3544, BW=1269MiB/s (1331MB/s)(223GiB/180010msec); 0 zone resets
    slat (usec): min=6, max=915863, avg=922.73, stdev=5002.70
    clat (msec): min=3, max=1021, avg=16.53, stdev=24.19
     lat (msec): min=3, max=1022, avg=17.45, stdev=25.06
    clat percentiles (msec):
     |  1.00th=[    9],  5.00th=[    9], 10.00th=[   10], 20.00th=[   11],
     | 30.00th=[   12], 40.00th=[   13], 50.00th=[   14], 60.00th=[   16],
     | 70.00th=[   17], 80.00th=[   18], 90.00th=[   21], 95.00th=[   23],
     | 99.00th=[   85], 99.50th=[  140], 99.90th=[  296], 99.95th=[  451],
     | 99.99th=[  860]
   bw (  MiB/s): min=   33, max= 1638, per=100.00%, avg=1276.91, stdev=107.54, samples=2142
   iops        : min=   94, max= 4730, avg=3565.55, stdev=225.36, samples=2142
  lat (msec)   : 4=0.01%, 10=10.32%, 20=46.63%, 50=40.48%, 100=1.42%
  lat (msec)   : 250=0.88%, 500=0.20%, 750=0.02%, 1000=0.04%, 2000=0.01%
  cpu          : usr=1.16%, sys=2.79%, ctx=2214429, majf=0, minf=10352
  IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=12.2%, 16=48.0%, 32=39.8%, >=64=0.0%
     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=0.1%, 16=0.1%, 32=0.1%, 64=0.0%, >=64=0.0%
     issued rwts: total=1327535,638073,0,0 short=0,0,0,0 dropped=0,0,0,0
     latency   : target=0, window=0, percentile=100.00%, depth=8

Run status group 0 (all jobs):
   READ: bw=2038MiB/s (2138MB/s), 2038MiB/s-2038MiB/s (2138MB/s-2138MB/s), io=358GiB (385GB), run=180010-180010msec
  WRITE: bw=1269MiB/s (1331MB/s), 1269MiB/s-1269MiB/s (1331MB/s-1331MB/s), io=223GiB (240GB), run=180010-180010msec

Disk stats (read/write):
  loop0: ios=1445458/638070, sectors=751500496/467976536, merge=0/0, ticks=15383195/7059551, in_queue=22569810, util=100.00%

VM results:
Code: 
random-read: (g=0): rw=randread, bs=(R) 256KiB-256KiB, (W) 256KiB-256KiB, (T) 256KiB-256KiB, ioengine=libaio, iodepth=32
...
random-write: (g=0): rw=randwrite, bs=(R) 256KiB-256KiB, (W) 256KiB-256KiB, (T) 256KiB-256KiB, ioengine=libaio, iodepth=16
...
mixed-readwrite: (g=0): rw=randrw, bs=(R) 128KiB-128KiB, (W) 128KiB-128KiB, (T) 128KiB-128KiB, ioengine=libaio, iodepth=24
...
sequential-read: (g=0): rw=read, bs=(R) 1024KiB-1024KiB, (W) 1024KiB-1024KiB, (T) 1024KiB-1024KiB, ioengine=libaio, iodepth=8
sequential-write: (g=0): rw=write, bs=(R) 1024KiB-1024KiB, (W) 1024KiB-1024KiB, (T) 1024KiB-1024KiB, ioengine=libaio, iodepth=8
fio-3.37
Starting 11 processes
random-read: Laying out IO file (1 file / 10240MiB)

random-read: (groupid=0, jobs=11): err= 0: pid=5769: Sat Jul 27 18:14:39 2024
  Description  : [Simulates reading pieces from different parts of files]
  read: IOPS=252, BW=62.1MiB/s (65.1MB/s)(10.9GiB/180570msec)
    slat (usec): min=4, max=1275.3k, avg=2148.51, stdev=36456.46
    clat (usec): min=176, max=3068.6k, avg=729867.16, stdev=356792.63
     lat (msec): min=11, max=3068, avg=732.02, stdev=354.75
    clat percentiles (msec):
     |  1.00th=[   93],  5.00th=[  211], 10.00th=[  292], 20.00th=[  414],
     | 30.00th=[  514], 40.00th=[  600], 50.00th=[  693], 60.00th=[  793],
     | 70.00th=[  902], 80.00th=[ 1028], 90.00th=[ 1200], 95.00th=[ 1351],
     | 99.00th=[ 1670], 99.50th=[ 1838], 99.90th=[ 2265], 99.95th=[ 2433],
     | 99.99th=[ 2970]
   bw (  KiB/s): min= 4864, max=225452, per=100.00%, avg=71179.55, stdev=5200.61, samples=2616
   iops        : min=   16, max=  766, avg=273.29, stdev=19.40, samples=2616
  write: IOPS=78, BW=22.4MiB/s (23.4MB/s)(4036MiB/180574msec); 0 zone resets
    slat (usec): min=7, max=1134.8k, avg=5045.61, stdev=54908.75
    clat (msec): min=16, max=2458, avg=795.13, stdev=365.74
     lat (msec): min=39, max=2458, avg=800.18, stdev=368.67
    clat percentiles (msec):
     |  1.00th=[  126],  5.00th=[  224], 10.00th=[  317], 20.00th=[  460],
     | 30.00th=[  575], 40.00th=[  684], 50.00th=[  785], 60.00th=[  885],
     | 70.00th=[  978], 80.00th=[ 1099], 90.00th=[ 1267], 95.00th=[ 1418],
     | 99.00th=[ 1754], 99.50th=[ 1854], 99.90th=[ 2140], 99.95th=[ 2265],
     | 99.99th=[ 2400]
   bw (  KiB/s): min= 3840, max=78710, per=100.00%, avg=31710.32, stdev=2369.85, samples=1765
   iops        : min=   12, max=  262, avg=95.45, stdev= 8.78, samples=1765
  lat (usec)   : 250=0.01%, 500=0.11%, 750=0.08%, 1000=0.01%
  lat (msec)   : 20=0.02%, 50=0.17%, 100=0.59%, 250=5.91%, 500=20.11%
  lat (msec)   : 750=26.61%, 1000=22.61%, 2000=23.55%, >=2000=0.23%
  cpu          : usr=0.02%, sys=0.06%, ctx=12126, majf=0, minf=10353
  IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=5.8%, 16=42.2%, 32=51.9%, >=64=0.0%
     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=0.1%, 16=0.1%, 32=0.1%, 64=0.0%, >=64=0.0%
     issued rwts: total=45594,14225,0,0 short=0,0,0,0 dropped=0,0,0,0
     latency   : target=0, window=0, percentile=100.00%, depth=8

Run status group 0 (all jobs):
   READ: bw=62.1MiB/s (65.1MB/s), 62.1MiB/s-62.1MiB/s (65.1MB/s-65.1MB/s), io=10.9GiB (11.8GB), run=180570-180570msec
  WRITE: bw=22.4MiB/s (23.4MB/s), 22.4MiB/s-22.4MiB/s (23.4MB/s-23.4MB/s), io=4036MiB (4232MB), run=180574-180574msec

Disk stats (read/write):
  vda: ios=46281/14508, sectors=22949376/8270400, merge=0/158, ticks=32239266/11054196, in_queue=43356955, util=99.27%
 
You must log in or register to reply here.
Top Bottom