I'm trying to figure out how exactly ESXi storage subsystem works. I'll give you an example.
The first peak [1] is the result of:
vmkfstools -i /vmfs/volumes/RAID0-1/333/333.vmdk /vmfs/volumes/RAID0-2/333/333
The second one [2] is the result of:
cp -r /vmfs/volumes/RAID0-1/333/ /vmfs/volumes/RAID0-2/333-dc/
The third [3] is the result of:
cat /vmfs/volumes/RAID0-1/random.100GiB >> /dev/null
where random.100GiB is the file created with
head -c 1073741824 < /dev/urandom > random.1GiBand concatenated 100 times.
Somehow vmkfstools utilizes disk at 100% level when copy command is just below 50% (and as you can see — with spikes).
Sending random file to /dev/null (as well as to another RAID0) is capped at 1Gbps (125 MBps) which is suspiciously the same as Ethernet speed.
When I use virtual disk in VM it looks like [1] therefore it also utilizes disk at 100%.
My setup is:
HP ProLiant DL180 w HP P840 RAID controller
Directly attached storages based on RAID-0.
This is not an error but I'm curious mastering my storage knowledge. Could someone explain me what's going on?