[fChk]

thanks fChk for your appreciation !
I used that dd command cause to me that seemed to be closer to the real / available / usable performance. I agree that it is usually the speed of write to write cache. But that is what we experience in our normal use.

using conv fdatasync asks the os to write from write cache to physical disk before dd exits. so here all the writes are first done to the cache and then purged to the disk. i guess it has a component of write to the disk time. so may be that's why some consider it as the measure of disk performance in a system. but in today's computers of huge cache and parallel access requests, cache purging to the disk time may not be that noticeable in real usage, unless the system is bogged down by write requests to overflow the cache...lol

oflag direct or dsync etc flag is kinda the synchronous access speed ....one may say the real speed of the disk.

Exactly!... I was convinced that you knew what you're doing!... But, disk performance tests should be about the disk not system caches; or at least showing both to get an idea of how the disk performs on its own and how much boost it gets when system caches enter the scene (as is the case in practice.)

Besides, presented as you did, people not paying much attention into those numbers meaning, get the wrong idea:

I think it is wonderful that the HDD from Manal gives such high performance, normally HDD has reasonable speeds but these speeds are indeed comparable with SSD for a 1gbps network uplink. Carry on @Manal!

i will do what you have requested when i get time and post for you. thanks again !!

No rush, take your time!