xfs
[Top] [All Lists]

Re: [PATCH v3 0/5] Rework mtime and ctime updates on mmaped writes

To: Jan Kara <jack@xxxxxxx>
Subject: Re: [PATCH v3 0/5] Rework mtime and ctime updates on mmaped writes
From: Andy Lutomirski <luto@xxxxxxxxxxxxxx>
Date: Wed, 4 Sep 2013 10:33:22 -0700
Cc: "linux-kernel@xxxxxxxxxxxxxxx" <linux-kernel@xxxxxxxxxxxxxxx>, "linux-ext4@xxxxxxxxxxxxxxx" <linux-ext4@xxxxxxxxxxxxxxx>, Dave Chinner <david@xxxxxxxxxxxxx>, "Theodore Ts'o" <tytso@xxxxxxx>, Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx>, xfs@xxxxxxxxxxx, Tim Chen <tim.c.chen@xxxxxxxxxxxxxxx>, Christoph Hellwig <hch@xxxxxxxxxxxxx>
Delivered-to: xfs@xxxxxxxxxxx
In-reply-to: <20130904150854.GF3996@xxxxxxxxxxxxx>
References: <cover.1377193658.git.luto@xxxxxxxxxxxxxx> <20130904150854.GF3996@xxxxxxxxxxxxx>
On Wed, Sep 4, 2013 at 8:08 AM, Jan Kara <jack@xxxxxxx> wrote:
> On Thu 22-08-13 17:03:16, Andy Lutomirski wrote:
>> Writes via mmap currently update mtime and ctime in ->page_mkwrite.
>> This hurts both throughput and latency.  In workloads that dirty a
>> large number of mmapped pages, ->page_mkwrite can be hot and
>> file_update_time is slow and scales poorly.  Updating timestamps can
>> also sleep, which hurts latency for real-time workloads.
>   It would help to make your case if you posted the latency comparison
> before & after the patchset in this introductory email. We can then see
> how significant is the reduction of latency...

Will do, although the data from my workload will be a little strange.

I was hoping that Dave Hansen would re-run his benchmark with these
patches applied.  I tried to run it, but it wasn't obvious what the
numbers that spewed out meant.

--Andy

>
>                                                                 Honza
> --
> Jan Kara <jack@xxxxxxx>
> SUSE Labs, CR



-- 
Andy Lutomirski
AMA Capital Management, LLC

<Prev in Thread] Current Thread [Next in Thread>