linux-ext4 - Re: [RFC] performance regression with "ext4: Allow parallel DIO reads"

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] [thread-next>] [day] [month] [year] [list]
Date:   Fri, 26 Jul 2019 19:57:29 -0600
From:   Andreas Dilger <adilger@...ger.ca>
To:     Joseph Qi <joseph.qi@...ux.alibaba.com>
Cc:     Joseph Qi <jiangqi903@...il.com>, Theodore Ts'o <tytso@....edu>,
        Jan Kara <jack@...e.cz>,
        Ext4 Developers List <linux-ext4@...r.kernel.org>,
        Xiaoguang Wang <xiaoguang.wang@...ux.alibaba.com>,
        Liu Bo <bo.liu@...ux.alibaba.com>
Subject: Re: [RFC] performance regression with "ext4: Allow parallel DIO reads"

It would be useful to post some details about your test hardware
(eg. HDD vs. SSD, CPU cores+speed, RAM), so that it is possible to make
a good comparison of someone sees different results. 

Cheers, Andreas

> On Jul 25, 2019, at 19:12, Joseph Qi <joseph.qi@...ux.alibaba.com> wrote:
> 
> 
> 
>> On 19/7/26 05:20, Andreas Dilger wrote:
>> 
>>> On Jul 23, 2019, at 5:17 AM, Joseph Qi <jiangqi903@...il.com> wrote:
>>> 
>>> Hi Ted & Jan,
>>> Could you please give your valuable comments?
>> 
>> It seems like the original patches should be reverted?  There is no data
> 
> From my test result, yes.
> I've also tested libaio with iodepth 16, it behaves the same. Here is the test
> data for libaio 4k randrw:
> 
> -------------------------------------------------------------------------------------------
> w/ parallel dio reads | READ 78313KB/s, 19578, 1698.70us  | WRITE 78313KB/s, 19578, 4837.60us
> -------------------------------------------------------------------------------------------
> w/o parallel dio reads| READ 387774KB/s, 96943, 1009.73us | WRITE 387656KB/s，96914, 308.87us
> -------------------------------------------------------------------------------------------
> 
> Since this commit went into upstream long time ago，to be precise, Linux
> 4.9, I wonder if someone else has also observed this regression, or
> anything I missed?
> 
> Thanks,
> Joseph
> 
>> in the original commit message that indicates there is an actual performance
>> improvement from that patch, but there is data here showing it hurts both
>> read and write performance quite significantly.
>>> Cheers, Andreas
>> 
>>>> On 19/7/19 17:22, Joseph Qi wrote:
>>>> Hi Ted & Jan,
>>>> I've observed an significant performance regression with the following
>>>> commit in my Intel P3600 NVMe SSD.
>>>> 16c54688592c ext4: Allow parallel DIO reads
>>>> 
>>>> From my initial investigation, it may be because of the
>>>> inode_lock_shared (down_read) consumes more than inode_lock (down_write)
>>>> in mixed random read write workload.
>>>> 
>>>> Here is my test result.
>>>> 
>>>> ioengine=psync
>>>> direct=1
>>>> rw=randrw
>>>> iodepth=1
>>>> numjobs=8
>>>> size=20G
>>>> runtime=600
>>>> 
>>>> w/ parallel dio reads : kernel 5.2.0
>>>> w/o parallel dio reads: kernel 5.2.0, then revert the following commits:
>>>> 1d39834fba99 ext4: remove EXT4_STATE_DIOREAD_LOCK flag (related)
>>>> e5465795cac4 ext4: fix off-by-one error when writing back pages before dio read (related)
>>>> 16c54688592c ext4: Allow parallel DIO reads
>>>> 
>>>> bs=4k:
>>>> -------------------------------------------------------------------------------------------
>>>> w/ parallel dio reads | READ 30898KB/s, 7724, 555.00us   | WRITE 30875KB/s, 7718, 479.70us
>>>> -------------------------------------------------------------------------------------------
>>>> w/o parallel dio reads| READ 117915KB/s, 29478, 248.18us | WRITE 117854KB/s，29463, 21.91us
>>>> -------------------------------------------------------------------------------------------
>>>> 
>>>> bs=16k:
>>>> -------------------------------------------------------------------------------------------
>>>> w/ parallel dio reads | READ 58961KB/s, 3685, 835.28us   | WRITE 58877KB/s, 3679, 1335.98us
>>>> -------------------------------------------------------------------------------------------
>>>> w/o parallel dio reads| READ 218409KB/s, 13650, 554.46us | WRITE 218257KB/s，13641, 29.22us
>>>> -------------------------------------------------------------------------------------------
>>>> 
>>>> bs=64k:
>>>> -------------------------------------------------------------------------------------------
>>>> w/ parallel dio reads | READ 119396KB/s, 1865, 1759.38us | WRITE 119159KB/s, 1861, 2532.26us
>>>> -------------------------------------------------------------------------------------------
>>>> w/o parallel dio reads| READ 422815KB/s, 6606, 1146.05us | WRITE 421619KB/s, 6587, 60.72us
>>>> -------------------------------------------------------------------------------------------
>>>> 
>>>> bs=512k:
>>>> -------------------------------------------------------------------------------------------
>>>> w/ parallel dio reads | READ 392973KB/s, 767, 5046.35us  | WRITE 393165KB/s, 767, 5359.86us
>>>> -------------------------------------------------------------------------------------------
>>>> w/o parallel dio reads| READ 590266KB/s, 1152, 4312.01us | WRITE 590554KB/s, 1153, 2606.82us
>>>> -------------------------------------------------------------------------------------------
>>>> 
>>>> bs=1M:
>>>> -------------------------------------------------------------------------------------------
>>>> w/ parallel dio reads | READ 487779KB/s, 476, 8058.55us  | WRITE 485592KB/s, 474, 8630.51us
>>>> -------------------------------------------------------------------------------------------
>>>> w/o parallel dio reads| READ 593927KB/s, 580, 7623.63us  | WRITE 591265KB/s, 577, 6163.42us
>>>> -------------------------------------------------------------------------------------------
>>>> 
>>>> Thanks,
>>>> Joseph
>>>> 
>> 
>> 
>> Cheers, Andreas
>> 
>> 
>> 
>> 
>>