Linux Storage Benchmark (IOPS)
-
iostat reports on disk usage, not disk capability. Very different things. You want to know the top speed of your disks, not how much they are being used. Like the difference between knowing your car can go 180 k/h, versus knowing it is currently going 60 k/h.
-
Start with hdparm. Something like this...
sudo hdparm -Tt yourdevice -
Or...
sudo hdparm -v yourdevice -
Is this a server or a desktop? If you have a GUI, check out gnome-disks
-
You can use dd, too. Mount the disk in question and do something like this...
dd if=/dev/zero of=/mydrivemount/output bs=8k count=10kJust delete that file when you are done.
-
FIO
-
II used hdparm, and this server or gui less, but I am not getting what I want, I want IOPS
like any windows utility notice the screenshot above SSD IOPS is ~55,000 R and 46000 Write
I want something that will allow me to get the IOPS in linux, those give me speed of file transfer but not IOPS
-
You'll need FIO as Pete says.
-
https://github.com/vladzcloudius/diskplorer
This is a cool wrapper for FIO, written by a colleague of mine. FIO provides you with the maximums, while this tool will allow you to measure the optimal settings and actual disk capabilities.
-
It is very good tool, but if you can tell him to also support write speeds not just read
-
@Emad-R said in Linux Storage Benchmark (IOPS):
It is very good tool, but if you can tell him to also support write speeds not just read
NVM I saw this -t option but
diskplorer.py: error: option -t: invalid choice: 'test' (choose from 'read', 'write', 'randwrite', 'randread')
However I dont get what IOPS in Windows tests and Linux tests, I am used to see IOPS like 100-200 for HDD and 40,000+ for SATA SSDs
~100,000 for NVMEhttps://www.samsung.com/semiconductor/minisite/ssd/product/consumer/850evo/
However it seems that maybe i dont know what I am looking for
RANDOM WRITE (4KB, QD1)
RANDOM WRITE (4KB, QD32)I just want one base metric to give me good indicator
FYi this is VM on HDD
-
@Emad-R IOPS varies wildly by how it is tested. The question for your Windows tests would be... how was it tested? We know how the Linux was tested, it tells us. Knowing how the Windows was tested is the real need at this point.
