I have a very old Linux server attached an iSCSI storage, I'm testing an iSCSI storage performance using
hdparm -tT /dev/devicename on this server. Then I discovered a phenomenon:
The iSCSI storages read/write performance will be slow than the server local HDD. But I know that the storages read/write performance should be much better than my old Linux server, then I am thinking that maybe the hdparm result are limited to local HDD performance, the result will not exceed local HDD's result.
Then I do some tests on 3 other servers to prove my point, and those tests results proved my point.
Seems it's clear for now, but I am asking my colleagues about this question, and he tells me that he disagreed, because the hdparm is running in the memory, so it's not limited to the local HDD performance, I thought it's make sense.
The hdparm results are limited to local HDD and/or memory? If so, in this case, the hdparm results are not the real speed of the storage device?