TRIM or no TRIM?

[Steven Hartland]

----- Original Message ----- 
From: "Mark Felder" <feld at freebsd.org>


> On Wed, Jul 24, 2013, at 11:08, Steven Hartland wrote:
>> 
>> Some people have said that, we've never seen it here, quite the opposite
>> in fact.
>> 
> 
> It's well documented. Here's a link to take you down the rabbit hole.
> 
> http://lists.freebsd.org/pipermail/freebsd-stable/2013-February/072171.html

I've seen that thread before, but it provides very little actual
evidence which would mean it effects all usages.

I even just tried the very test listed in Jeremy's post and I'm
unable to reproduce even the slightest sutter, let alone a total
I/O stall.
http://lists.freebsd.org/pipermail/freebsd-fs/2011-October/012752.html

Just to be clear I'm not saying there's no issue, just we've never
experienced it and we use compression on a heavily used MySQL DB,
our RRD processing machine and hundreds of game servers (which are
highly latency sensitive).

Clearly if your performing an additional translation of the data
theirs going to be an CPU overhead for doing so but if said overhead
is smaller that the gain in the reduction of data written or read 
from the device then your still going to end up with better performance
with compression than without.

Maybe we're lucky and our data matches the magic pattern or perhaps
others are exadurating the negative effect, possibly a combination
of the two I don't know ;-)

    Regards
    Steve

================================================
This e.mail is private and confidential between Multiplay (UK) Ltd. and the person or entity to whom it is addressed. In the event of misdirection, the recipient is prohibited from using, copying, printing or otherwise disseminating it or any information contained in it. 

In the event of misdirection, illegible or incomplete transmission please telephone +44 845 868 1337
or return the E.mail to postmaster at multiplay.co.uk.