1

How many milliseconds of await is considered to be high?

i have 800ms and 600ms

iostat -x

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
           1.28    0.00    0.22    0.90    0.00   97.60

Device:         rrqm/s   wrqm/s   r/s   w/s   rsec/s   wsec/s avgrq-sz avgqu-sz   await  svctm  %util
cciss/c0d0        0.14    29.17  0.50 36.79    53.47   528.92    15.62     1.22   32.80   3.29  12.28
cciss/c0d0p1      0.00     0.00  0.00  0.00     0.01     0.00    19.23     0.00   12.43   8.85   0.00
cciss/c0d0p2      0.02    22.06  0.07 10.28     2.35   259.12    25.25     0.70   67.52   2.35   2.43
cciss/c0d0p3      0.00     0.00  0.00  0.00     0.01     0.00    36.00     0.00    4.07   4.07   0.00
cciss/c0d0p4      0.12     7.11  0.43 26.51    51.11   269.80    11.92     0.52   19.44   4.21  11.33
sdb               0.00     0.00  0.00  0.00     0.01     0.00    10.69     0.00   77.13  29.98   0.00
sdc               0.00     0.00  0.00  0.00     0.13     0.00   172.12     0.00  868.05 862.83   0.07
sdd               0.00     0.00  0.00  0.00     0.02     0.00    20.58     0.00  632.25 242.49   0.02
Gabriel Sousa
  • 317
  • 2
  • 4
  • 10
  • 1
    For comparison I am seeing `11.95` and `6.70` on the two disks in a RAID-1, which is a performance bottleneck I am currently looking for a solution to. – kasperd Sep 19 '15 at 07:34

1 Answers1

3

"High" is entirely subjective. Any metric is "high" when it is causing a degradation in service quality beyond what is acceptable to whomever you are providing the service to.

womble
  • 96,255
  • 29
  • 175
  • 230