[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <1374108910-50444-1-git-send-email-tmac@hp.com>
Date:	Wed, 17 Jul 2013 18:55:10 -0600
From:	T Makphaibulchoke <tmac@...com>
To:	viro@...iv.linux.org.uk, tytso@....edu, adilger.kernel@...ger.ca,
	linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org,
	linux-ext4@...r.kernel.org, tmakph@...il.com
Cc:	aswin@...com
Subject: [PATCH 0/2] ext4: increase mbcache scalability
This patch intends to improve the scalability of an ext4 filesystem by
introducing higher degree of parallelism to the usages of its mb_cache and
mb_cache_entries.
Here are some of the benchmark results with the changes. 
On a 90 core machine:
Here are the performance improvements in some of the aim7 workloads,
---------------------------
|             | % increase |
---------------------------
| custom      |     13.30  |
---------------------------
| disk        |      5.00  |  
---------------------------
| new_fserver |      6.73  |  
---------------------------
| shared      |      7.12  |  
---------------------------
For Swingbench dss workload,
-------------------------------------------------------------------------------
| Users        | 100  | 200  | 300  | 400  | 500  | 600  | 700  | 800  | 900  |
-------------------------------------------------------------------------------
| % imprvoment | 9.14 | 9.26 | 19.9 | 5.31 | 7.79 | 0.23 |-4.39 |-4.48 |-4.36 |
-------------------------------------------------------------------------------
For SPECjbb2013, composite run,
--------------------------------------------
|               | max-jOPS | critical-jOPS |
--------------------------------------------
| % improvement |   6.73   |     16.37     |
--------------------------------------------
On an 80 core machine:
Here are the improvments in some of the aim7 workloads,
 
---------------------------
|             | % increase |
---------------------------
| fserver     |      6.47  |
---------------------------
| new_fserver |      4.62  |  
---------------------------
For Swingbench dss workload,
-------------------------------------------------------------------------------
| Users        | 100  | 200  | 300  | 400  | 500  | 600  | 700  | 800  | 900  |
-------------------------------------------------------------------------------
| % imprvoment | 4.29 | 2.28 | 3.25 | 1.45 |-0.84 | 3.11 |-0.61 | 3.80 | 2.58 |
-------------------------------------------------------------------------------
For SPECjbb2013, composite run,
--------------------------------------------
|               | max-jOPS | critical-jOPS |
--------------------------------------------
| % improvement |   0.49   |      3.21     |
--------------------------------------------
The changes have been tested with ext4 xfstests to verify that no regression
has been introduced.  
 fs/ext4/ext4.h          |   1 +
 fs/ext4/super.c         |   8 ++
 fs/ext4/xattr.c         |  41 +++++---
 fs/ext4/xattr.h         |   3 +
 fs/mbcache.c            | 432 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++------------------
 include/linux/mbcache.h |   5 +
 6 files changed, 385 insertions(+), 105 deletions(-)
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/
Powered by blists - more mailing lists
 
