Re: Benchmarking: How to identify bottleneck (limiting factor) and achieve "linear scalability"?

Le 29/01/2019 à 07:15, Saurabh Nanda a écrit :
>
> c) I tried another cloud hosting provider (E2E Networks) and just
> the raw performance numbers (with default configuration) are
> blowing Hetzner out of the water.
>
>
> I noticed that on E2E, the root filesystem is mounted with the
> following options:
>
>     /dev/xvda on / type ext4
> (rw,noatime,nodiratime,nobarrier,errors=remount-ro,stripe=512,data=ordered)
>
> whereas on Hetzner, it is mounted with the following options:
>
>     /dev/nvme0n1p3 on / type ext4
>     (rw,relatime,data=ordered)
>
> How much of a difference can this have on absolute TPS numbers?
>

Differences can be significative. noatime does not update inode access
time, while relatime updates the inode access time if the change time
was before access time (which can be often the case for a database)

nobarrier disable block-level write barriers. Barriers ensure that data
is effectively stored on system, The man command says: "If disabled  on 
a  device with  a  volatile (non-battery-backed)  write-back  cache, 
the nobarrier option will lead to filesystem corruption on a  system
crash or power loss."

You should probably consider noatime compared to relatime, and
nobarriers depends if you have a battery or not

Also, this is an SSD, so you should TRIM  it, either with preiodical
fstrim, or using the discard option

Nicolas