In the past I’ve never noticed much gain from going to a fixed disk over dynamic so I decided to finally put it to the test. I tested with two different disk arrays, one a slow SATA using storage spaces (RAID1+0) and another using SSD in straight RAID 0. Each array has 4 disks. Each connected over SATA600. The speed of the disks ultimately doesn’t matter, I just want to find out if fixed really is faster than dynamic. I’m using crystal disk mark 4.0.1 x64 inside a VM. I ran two tests each, 3 passes with a 1GB file.
I first tested with SSD and after the results I thought maybe it was just SSD since those disks are so fast anyway. Maybe no latency wipes out any performance loss you get with the 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 is awesome, incredibly high. But those numbers are consistent in SSD as well so I don’t think it’s a 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]
I also did some more testing with Atto and got the same results. In each case Dynamic was always faster than fixed. I’m confused …
Dynamic test
Fixed test
To be honest, they are pretty much even in performance, aside from those crazy read numbers. But it’s also clear that dynamic edges out fixed. Maybe it’s my environment or something else, because everything online says Fixed should be faster. There are even other bench marks.