On Tue, Jun 7, 2016 at 4:47 PM, Jens Axboe <axboe@xxxxxxxxx> wrote: > On 06/06/2016 03:41 PM, Karan Singh wrote: >> >> #1 : Is r/s and w/s fields in iostat output corresponds to read and >> write IOPS ? ( >> https://serverfault.com/questions/342273/disk-iops-count-in-gnu-linux >> mentions r/s and w/s as IOPS) >> #2 : If yes then why does they dont match with FIO IOPS results ? > > > The answer is no, and that goes for any application. You could have an app > that does sequential ios in 4k chunks, and the kernel merges them into 512k > IOs. From the application perspective, it's doing 128 as many iops as the > device is. The converse can also be true, like in your case, since the > application is submitting IO that is larger than the device can handle. > Hence it's broken up, and the result is that the device is now doing a lot > more IOPS than the application is issuing. Thanks Jens for answering You said kernel merges / divides IO into 512k chunk. #1 How can i verify this (512k) value, any command or file to look this into ? #2 Out of curiosity is there a way to disable this merging / division of IO ? it might be not efficient but just want to play around this. #3 Is it true that merging /division of IO is irrespective of pattern ( sequential or random ) i.e kernel will perform this whether the IO is seq. or random ? > > -- > Jens Axboe > -- To unsubscribe from this list: send the line "unsubscribe fio" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
