Measuring disk latency by writing/reading a file

Discussion created by Robert_Borowicz on Feb 28, 2015
Latest reply on Feb 28, 2015 by Garin

I have this burning desire to create a threshold measurement that measures real user pain when disk latency is pronounced. I want to take a file of say 1's and 0's and write it to each drive (in windows) and partition in Unix/Linux and measure how long it takes. When it surpasses a threshold throw an alarm.


Is this possible with any probe?  I'm new and just now examining iostat.


In our case (Solaris) I think I/O has become conjested system wide, while *individual* disk partitions are performing reasonably accoring to Solaris iostat. Not sure I have a sound theory and doing an actual file W/R to me would be one way to prove that point.


Anybody done such a thing?


Is there a probe that will let you execute a cmd line snippet, and measure the time it takes?


Thanks in advance,