Fixed vs Dynamic Benchmarks, fixed is slower

in past i've never noticed gain going fixed disk on dynamic decided put test. tested 2 different disk arrays, 1 slow sata using storage spaces (raid1+0) , using ssd in straight raid 0. each array has 4 disks. each connected on sata600. speed of disks doesn't matter, want find out if fixed faster dynamic. i'm using crystal disk mark 4.0.1 x64 inside vm. ran 2 tests each, 3 passes 1gb file. on 2012 r2 , vhdx. 

i first tested ssd , after results thought maybe ssd since disks fast anyway. maybe no latency wipes out performance loss overhead of dynamic.

dynamic test 1
   sequential read (q= 32,t= 1) :  5568.709 mb/s
  sequential write (q= 32,t= 1) :   662.037 mb/s
  random read 4kib (q= 32,t= 1) :   139.494 mb/s [ 34056.2 iops]
 random write 4kib (q= 32,t= 1) :   123.682 mb/s [ 30195.8 iops]
         sequential read (t= 1) :   579.854 mb/s
        sequential write (t= 1) :   576.732 mb/s
   random read 4kib (q= 1,t= 1) :    19.219 mb/s [  4692.1 iops]
  random write 4kib (q= 1,t= 1) :    34.639 mb/s [  8456.8 iops]

dynamic test 2
   sequential read (q= 32,t= 1) :  4511.435 mb/s
  sequential write (q= 32,t= 1) :   662.157 mb/s
  random read 4kib (q= 32,t= 1) :   140.051 mb/s [ 34192.1 iops]
 random write 4kib (q= 32,t= 1) :   123.475 mb/s [ 30145.3 iops]
         sequential read (t= 1) :   576.089 mb/s
        sequential write (t= 1) :   575.869 mb/s
   random read 4kib (q= 1,t= 1) :    19.542 mb/s [  4771.0 iops]
  random write 4kib (q= 1,t= 1) :    33.126 mb/s [  8087.4 iops]

fixed test 1
   sequential read (q= 32,t= 1) :   700.012 mb/s
  sequential write (q= 32,t= 1) :   659.203 mb/s
  random read 4kib (q= 32,t= 1) :   140.867 mb/s [ 34391.4 iops]
 random write 4kib (q= 32,t= 1) :   124.820 mb/s [ 30473.6 iops]
         sequential read (t= 1) :   570.374 mb/s
        sequential write (t= 1) :   570.917 mb/s
   random read 4kib (q= 1,t= 1) :    18.859 mb/s [  4604.2 iops]
  random write 4kib (q= 1,t= 1) :    34.872 mb/s [  8513.7 iops]

fixed test 2
   sequential read (q= 32,t= 1) :   699.533 mb/s
  sequential write (q= 32,t= 1) :   656.096 mb/s
  random read 4kib (q= 32,t= 1) :   141.714 mb/s [ 34598.1 iops]
 random write 4kib (q= 32,t= 1) :   125.472 mb/s [ 30632.8 iops]
         sequential read (t= 1) :   584.835 mb/s
        sequential write (t= 1) :   568.106 mb/s
   random read 4kib (q= 1,t= 1) :    18.880 mb/s [  4609.4 iops]
  random write 4kib (q= 1,t= 1) :    35.024 mb/s [  8550.8 iops]

slower sata now. read awesome, incredibly high. numbers consistent in ssd don't think it's fluke. 

dynamic test 1
   sequential read (q= 32,t= 1) :  1768.289 mb/s
  sequential write (q= 32,t= 1) :    25.291 mb/s
  random read 4kib (q= 32,t= 1) :     6.339 mb/s [  1547.6 iops]
 random write 4kib (q= 32,t= 1) :     0.304 mb/s [    74.2 iops]
         sequential read (t= 1) :    67.744 mb/s
        sequential write (t= 1) :    26.213 mb/s
   random read 4kib (q= 1,t= 1) :     0.593 mb/s [   144.8 iops]
  random write 4kib (q= 1,t= 1) :     0.310 mb/s [    75.7 iops]

dynamic test 2
   sequential read (q= 32,t= 1) :  1723.136 mb/s
  sequential write (q= 32,t= 1) :    25.242 mb/s
  random read 4kib (q= 32,t= 1) :     6.186 mb/s [  1510.3 iops]
 random write 4kib (q= 32,t= 1) :     0.315 mb/s [    76.9 iops]
         sequential read (t= 1) :    66.275 mb/s
        sequential write (t= 1) :    25.581 mb/s
   random read 4kib (q= 1,t= 1) :     0.565 mb/s [   137.9 iops]
  random write 4kib (q= 1,t= 1) :     0.312 mb/s [    76.2 iops]

fixed test 1
   sequential read (q= 32,t= 1) :    67.003 mb/s
  sequential write (q= 32,t= 1) :    24.195 mb/s
  random read 4kib (q= 32,t= 1) :     0.658 mb/s [   160.6 iops]
 random write 4kib (q= 32,t= 1) :     0.264 mb/s [    64.5 iops]
         sequential read (t= 1) :    65.013 mb/s
        sequential write (t= 1) :    25.376 mb/s
   random read 4kib (q= 1,t= 1) :     0.607 mb/s [   148.2 iops]
  random write 4kib (q= 1,t= 1) :     0.263 mb/s [    64.2 iops]

fixed test 2
   sequential read (q= 32,t= 1) :    77.594 mb/s
  sequential write (q= 32,t= 1) :    24.769 mb/s
  random read 4kib (q= 32,t= 1) :     0.636 mb/s [   155.3 iops]
 random write 4kib (q= 32,t= 1) :     0.261 mb/s [    63.7 iops]
         sequential read (t= 1) :    65.016 mb/s
        sequential write (t= 1) :    25.586 mb/s
   random read 4kib (q= 1,t= 1) :     0.614 mb/s [   149.9 iops]
  random write 4kib (q= 1,t= 1) :     0.269 mb/s [    65.7 iops]

so why isn't fixed disk giving performance better? writes should @ least better right? , i'm not sure why reads high on slow sata disks dynamic. local storage no csv cache. 

actually of 2012 release (if recall features correctly) performance window between virtual disk types closed.

at time expect no significant different between dynamic or fixed virtual disks.  (until depths of fragmentation).

this based on introduction of vhdx.  (which way backs storage spaces).

here old reference server 2008 r2 when began closing performance difference window: http://download.microsoft.com/download/0/7/7/0778c0bb-5281-4390-92cd-ec138a18f2f9/ws08_r2_vhd_performance_whitepaper.docx

and vhdx specification announcement: http://blogs.msdn.com/b/virtual_pc_guy/archive/2012/09/27/vhdx-specification-available.aspx introduced close of performance differences.

---

brian ehlert
http://itproctology.blogspot.com
learn. apply. repeat.

Windows Server  >  Hyper-V