How a Linux server (dedicated or a VPS) admin can discover his/her server disk/drive/HDD/SSD performance in the IOPS/tps (transactions/operations per second)?
So far this command shows numbers that can be related to real IOPS:
(requires "fio" be installed - yum install fio/apt-get install fio)echo "Can take a few seconds..";fio --bs=4k --size=1m --direct=1 --runtime=5 --rw=randread --numjobs=3 --group_reporting --name=1m|grep iops|awk '{print $5}';rm -f 1m.*
one can add it into while loop so it continuously show current reading IOPS of the fio test:
iops=768while true;do sleep 1 && fio --bs=4k --size=1m --direct=1 --runtime=5 --rw=randread --numjobs=3 --group_reporting --name=1m|grep iops|awk '{print $5}' && rm -f 1m.*;done
iops=5907
iops=3522
iops=5605
iops=1252
iops=3428
iops=33391
iops=33391
When value is low, it can mean other process is using IO i assume.
On SSD RAID 10 (6xDELL SSD) idle i got roughly values around 2000 :-/
Above mentioned command is also mentioned in this topic, and i am writing this as a newbie, if anyone can narrow this or come with more accurate command, please kindly share it.
UPDATE: i think this IOPS test can have better accuracy:
Example result at https://internetlifeforum.com/vps-hosting/9431-wedos-review-vpson/while true;do sleep 1 && fio --bs=4k --size=1m --direct=1 --runtime=5 --rw=randread --numjobs=3 --group_reporting --name=1m|grep iops|awk '{print $5}' && rm -f 1m.*;done
Bookmarks