write block size - ssd

Hello

I have a problem with understanding write block size.
Writes on ssd volume, type is ext4 block size to 4096, lsblk (PHY-SEC) for volume gives me 512

I run
dd if=/dev/urandom of=output2 obs=512000 conv=fsync count=10000000
and after that i run
dd if=/dev/urandom of=output2 obs=512 conv=fsync count=100000000

and the graph is identical.
I expected significant differences, what i miss ?

How to simulate change write block size ?






In some situations on real data, write block size above 150KB increases my response write even above 10ms.

Examples from dd want to understand what is going on.





Comments

  • Hi,

    block size is determined by: data rate/IOPS

    Do you see increase of data rate and IOPS during on your volume during your test?

  • I didn't know that, thank you

    I thought it was layer 7, application in different block size gives data to write.
    But then I didn't understand the block size on the file system, which doesn't change.

    Thank you for explanation






Sign In or Register to comment.