[svlug] SSD was: Slides from the GoLUG meeting

Rick Moen rick at linuxmafia.com
Fri Jun 3 10:48:21 PDT 2016


Quoting Ivan Sergio Borgonovo (mail at webthatworks.it):

> I think deciding what's best is *now* more complicated than "scheduler A 
> is simpler and it looks more adequate".

Isn't that mostly a matter of doing a modest amount of local testing and
seeing if there's a performance difference?  I mean, my basic point
during my presentation and later has been that it's easy at runtime to
switch schedulers so you can try what's available.  Then, when you
determine what's best, through local usage, you make your choice
persistent by passing that to the kernel at boot time.

> I didn't find many recent benchmark comparing cfq to deadline since 
> there is a new scheduler that is attracting attentions (bfq) but reading
> 
> https://blog.pgaddict.com/posts/postgresql-io-schedulers-cfq-noop-deadline
> 
> it looks as it is at least debatable which scheduler to chose.
> 
> I'm just suggesting that if anyone really really need to know should do 
> his homework again.

Always fair and good advice.

Referenced article by the pgaddict.com guy (Tomas Vondra) begins with an
important truth in this regard:  '[I]n my experience, many recommended
settings are effectively myths.'  True.  Moreover, things true in other
people's use-cases may not be true in yours.  However, as I reminded 
Wednesday's audience, it's also common to fool one's self with
benchmarking -- thus in the case of dd, my recommending a basic guide
(https://romanrm.net/dd-benchmark) so they wouldn't accidentally
benchmark RAM caching.

Anyway, thanks for the Vondra article.  Seems meaty, and I'm working
through it.




More information about the svlug mailing list