linux-ext4 - Re: What should be the optimum value of G (number-of-groups) when using flex

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-Id: <701D25F0-1D33-420C-8D5C-C65734E32E13@dilger.ca>
Date:	Wed, 7 Nov 2012 05:02:25 -0700
From:	Andreas Dilger <adilger@...ger.ca>
To:	Ashish Sangwan <ashishsangwan2@...il.com>
Cc:	"Theodore Ts'o" <tytso@....edu>,
	Namjae Jeon <linkinjeon@...il.com>, linux-ext4@...r.kernel.org
Subject: Re: What should be the optimum value of G (number-of-groups) when using flex_bg?

On 2012-11-05, at 9:36 PM, Ashish Sangwan wrote:
> We have a 2TB HDD which we formated using default options of mkfs.ext4(v1.42.6).
> I took around ~15 seconds to complete this operation.
> 
> Than we formatted the same partition with "-G 1024" and there was
> around 30% performance improvement in
> the mkfs.ext4 timings.
> 
> Than, out of curiosity, we increased the G's value to 1048576 and
> there was a drastic degradation in the performance of mkfs.ext4.
> It took more than 5 minutes to complete the execution.
> 
> Is it the expected behavior?

The optimum number is very workload specific.  We use "-G 256" and
a few other options so that the number of metadata blocks per group
is a multiple of 1MB and aligned on 1MB boundaries.  This is for
optimized IO on RAID-6 storage that has good IO performance with 1MB
chunk sizes.

> AFAIK flex_bg is on by default with G's vaule as 16. Can we consider
> this the optimum value?
> 
> Any suggesstions on what  we can set the Gs value for a 2TB HDD such
> that both the FS and the
> mkfs.ext4 gives optimum performance?
> It would be really helpful if somebody can shed some light over this feature.
> 
> Thanks,
> Ashish


Cheers, Andreas





--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html