LDP
/
old-www
1
1
Fork 0
Code Issues Pull Requests Releases Wiki Activity
old-www/LDP/LG/issue68/dellomodarme.html

1333 lines
31 KiB
HTML
Raw Permalink Blame History

<!--startcut ==============================================-->
<!-- *** BEGIN HTML header *** -->
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML><HEAD>
<title>Journalling Filesystems for Linux LG #68</title>
</HEAD>
<BODY BGCOLOR="#FFFFFF" TEXT="#000000" LINK="#0000FF" VLINK="#0000AF"
ALINK="#FF0000">
<!-- *** END HTML header *** -->
<CENTER>
<A HREF="http://www.linuxgazette.com/">
<IMG ALT="LINUX GAZETTE" SRC="../gx/lglogo.png"
WIDTH="600" HEIGHT="124" border="0"></A>
<BR>
<!-- *** BEGIN navbar *** -->
<IMG ALT="" SRC="../gx/navbar/left.jpg" WIDTH="14" HEIGHT="45" BORDER="0" ALIGN="bottom"><A HREF="collinge.html"><IMG ALT="[ Prev ]" SRC="../gx/navbar/prev.jpg" WIDTH="16" HEIGHT="45" BORDER="0" ALIGN="bottom"></A><A HREF="index.html"><IMG ALT="[ Table of Contents ]" SRC="../gx/navbar/toc.jpg" WIDTH="220" HEIGHT="45" BORDER="0" ALIGN="bottom" ></A><A HREF="../index.html"><IMG ALT="[ Front Page ]" SRC="../gx/navbar/frontpage.jpg" WIDTH="137" HEIGHT="45" BORDER="0" ALIGN="bottom"></A><A HREF="http://www.linuxgazette.com/cgi-bin/talkback/all.py?site=LG&article=http://www.linuxgazette.com/issue68/dellomodarme.html"><IMG ALT="[ Talkback ]" SRC="../gx/navbar/talkback.jpg" WIDTH="121" HEIGHT="45" BORDER="0" ALIGN="bottom" ></A><A HREF="../faq/index.html"><IMG ALT="[ FAQ ]" SRC="./../gx/navbar/faq.jpg"WIDTH="62" HEIGHT="45" BORDER="0" ALIGN="bottom"></A><A HREF="ghosh.html"><IMG ALT="[ Next ]" SRC="../gx/navbar/next.jpg" WIDTH="15" HEIGHT="45" BORDER="0" ALIGN="bottom" ></A><IMG ALT="" SRC="../gx/navbar/right.jpg" WIDTH="15" HEIGHT="45" ALIGN="bottom">
<!-- *** END navbar *** -->
<P>
</CENTER>
<!--endcut ============================================================-->
<H4 ALIGN="center">
"Linux Gazette...<I>making Linux just a little more fun!</I>"
</H4>
<P> <HR> <P>
<!--===================================================================-->
<center>
<H1><font color="maroon">Journalling Filesystems for Linux</font></H1>
<H4>By <a href="mailto:matt@martine2.difi.unipi.it">Matteo Dell'Omodarme</a></H4>
</center>
<P> <HR> <P>
<!-- END header -->
<h1 align=center>Introduction</h1>
A filesystem is the software used to organize and manage
the data stored on disk drives; it ensures the
integrity of the data providing that data written
to disk is identical when it is read
back. In addition to storing the data contained in files, a filesystem
also stores and manages important
information about the files and about the filesystem
itself (i.e. date and time stamps, ownership, access permissions,
the file's size and the storage location or locations on
disk, and so on). This information is commonly referred to as metadata. <p>
Since a filesystem tries to work as asynchronous as possible, in order to
avoid hard-disk bottleneck, a sudden interruption of its
work could result in a loss of data.
As an example, let's consider the following scenario: what happens if your machine crashes
when you are working on a document residing on a Linux standard ext2
filesystem? <br>
There are several answers:
<ul>
<li> The machine crashes after you saved the file. This is the best
scenario: you haven't lost anything. Just reboot the machine and
continue working on the document. <P>
<li>The machine crashes before you saved the file.
You have lost all your changes but your old version is still ok. <P>
<li>The machine crashes during the
exact moment when the file is being written.
This is the worst case: the new version of the file is
physically overwriting the old version. You end up with a file
partially new and partially old. If the file was written in a binary form
you can't reopen it because the internal format of
its data is inconsistent with what the application expects. <P>
</ul>
In this last scenario things can be even worse if the drive was
writing the metadata areas, such
as the directory itself. Now instead of one corrupted file, you have one
corrupted filesystem and you can lose an entire directory
or all the data on an entire disk partition.<p>
The standard Linux filesystem (ext2fs) makes an attempt to prevent
and recover from the metadata corruption case performing an extensive
filesystem analysis (fsck) during bootup.
Since ext2fs incorporates redundant copies of critical
metadata, it is extremely unlikely for that data to be completely
lost. The system figures out where the corrupt metadata is, and then
either repairs the damage by copying from the redundant version or
simply deletes the file or files whose metadata is affected. <p>
Obviously, the larger is the filesystem to check, the longer
the check process. On a partition of several gigabytes it may
take a great deal of time to check the metadata during bootup.<br>
As
Linux begins to take on more complex applications, on larger servers,
and with less tolerance for downtime, there is a need for more
sophisticated filesystems that do an even better job of protecting data
and metadata.<p>
The journalling filesystems available for Linux are the
answer to this need.
<h1 align=center>What is a journalling filesystem?</h1>
Here is reported only a general introduction to journalling. For more
specific and technical notes please see Juan I. Santos Florido
article in <a
href="../issue55/florido.html">Linux Gazette 55</a>.
Other information can be obtained from
<a href="http://freshmeat.net/articles/view/212/">freshmeat.net/articles/view/212/</a>.<p>
Most modern filesystems use journalling techniques borrowed from the
database world to improve crash
recovery. Disk transactions are written sequentially to an area of
disk called <em>journal</em> or <em>log</em> before being
written to their final locations within the filesystem. <br>
Implementations vary in terms of what data is
written to the log. Some implementations write only the filesystem
metadata, while others record all writes to the journal.<p>
Now, if a crash happens before
the journal entry is committed, then the original data is still on the
disk and you lost only
your new changes.
If the
crash happens during the actual disk update (i.e. after the journal
entry was committed), the journal
entry shows what was supposed to have happened. So when the
system reboots, it can simply replay the journal entries and complete the
update that was interrupted. <p>
In either case, you have valid data and not a trashed partition.
And since the recovery time associated with this log-based approach is
much shorter, the system is on line in few seconds.<p>
It is also important to note that using a journalling filesystem does not
entirely obsolete the use of filesystem
checking programs (fsck). Hardware and software errors that corrupt random
blocks in the filesystem are not
generally recoverable with the transaction log.
<h1 align=center>Available journalling filesystems</h1>
In the following part I will consider three journalling filesystems.<p>
The first one is <b>ext3</b>. Developed by Stephen Tweedie, a leading
Linux kernel developer, ext3 adds journalling into ext2.
It is available in alpha
form at
<a href="ftp://ftp.linux.org.uk/pub/linux/sct/fs/jfs/">ftp.linux.org.uk/pub/linux/sct/fs/jfs/</a>.<p>
Namesys has a journalling filesystem under development called
<b>ReiserFS</b>. It is available at
<a href="http://www.namesys.com/">www.namesys.com</a>.<p>
SGI has released on May 1 2001 version 1.0 of its <b>XFS</b> filesystem for
Linux. You can find it at
<a href="oss.sgi.com/projects/xfs/">oss.sgi.com/projects/xfs/</a>.<p>
In this article these three solutions are tested and benchmarked using
two different programs. <p>
<h1 align=center>Installing ext3</h1>
For technical notes about ext3 filesystem please refer to Dr.
Stephen Tweedie's <a
href="ftp://ftp.kernel.org/pub/linux/kernel/people/sct/ext3/">paper</a>
and to his
<a href=http://olstrans.sourceforge.net/release/OLS2000-ext3/OLS2000-ext3.html">talk</a>.
<p>
The ext3 filesystem is directly derived from its ancestor, ext2. It has
the valuable characteristic to be absolutely backward compatible to
ext2 since it is just an ext2 filesystem with journalling.
The obvious drawback is that ext3 doesn't implement any of the modern
filesystem features which increase data manipulation speed and packing. <p>
ext3 comes as a patch of
2.2.19 kernel, so first of all, get a linux-2.2.19 kernel from
<a href="ftp://ftp.kernel.org/pub/linux/kernel/v2.2/">ftp.kernel.org</a>
or from one of its <a href="http://www.kernel.org/mirrors/">mirrors</a>.
The patch is available at <a
href="ftp://ftp.linux.org.uk/pub/linux/sct/fs/jfs/">ftp.linux.org.uk/pub/linux/sct/fs/jfs</a>
or <a
href="ftp://ftp.kernel.org/pub/linux/kernel/people/sct/ext3">ftp.kernel.org/pub/linux/kernel/people/sct/ext3</a>
or from one mirror of this site.<br>
From one of these sites you need to get the following files:
<ul>
<li><a
href="ftp://ftp.kernel.org/pub/linux/kernel/people/sct/ext3/ext3-0.0.7a.tar.bz2">ext3-0.0.7a.tar.bz2</a>:
the kernel patch.
<li><a
href="ftp://ftp.kernel.org/pub/linux/kernel/people/sct/ext3/e2fsprogs-1.21-WIP-0601.tar.bz2">e2fsprogs-1.21-WIP-0601.tar.bz2</a>:
the e2fsprogs suite with ext3 support.
</ul>
Copy Linux kernel linux-2.2.19.tar.bz2 and ext3-0.0.7a.tar.bz2 files to
<em>/usr/src</em> directory and extract them:
<pre>
mv linux linux-old
tar -Ixvf linux-2.2.19.tar.bz2
tar -Ixvf ext3-0.0.7a.tar.bz2
cd linux
cat ../ext3-0.0.7a/linux-2.2.19.kdb.diff | patch -sp1
cat ../ext3-0.0.7a/linux-2.2.19.ext3.diff | patch -sp1
</pre>
The first diff is copy of SGI's kdb kernel debugger patches. The
second one is the ext3 filesystem. <br>
Now, configure the kernel, saying YES to "Enable Second extended fs
development code" in the filesystem section, and build it.<p>
After the kernel is compiled and installed you should make and install
the e2fsprogs:
<pre>
tar -Ixvf e2fsprogs-1.21-WIP-0601.tar.bz2
cd e2fsprogs-1.21
./configure
make
make check
make install
</pre>
That's all. The next step is to make an ext3 filesystem in a partition.
Reboot with the new kernel. Now you have two options: make a
new journalling filesystem or journal an existing one. <br>
<ul>
<li>Making a new ext3 filesystem. Just use the mke2fs from the
installed e2fsprogs, and use the "-j" option when running mke2fs:
<pre>
mke2fs -j /dev/xxx
</pre>
where /dev/xxx is the device where you would create the ext3 filesystem.
The "-j" flag tells mke2fs to create an ext3 filesystem with a hidden
journal. You could control the size of the journal using the optional
flag -Jsize=&lt;n&gt; (n is the preferred size of the journal in Mb).
<li>Upgrade an existing ext2 filesystem to ext3. Just use tune2fs:
<pre>
tune2fs -j /dev/xxx
</pre>
You should do that either on mounted or unmounted filesystem. If the
filesystem is mounted a file .journal is created in the top-level
directory of the filesystem; if it is unmounted a hidden system inode
is used for the journal. In such a way all the data in the filesystem
are preserved.
</ul>
You can mount the ext3 filesystem using the command:
<pre>
mount -t ext3 /dev/xxx /mount_dir
</pre>
Since ext3 is basically ext2 with journalling, a cleanly unmounted ext3
filesystem could be remounted as ext2 without any other commands.
<h1 align=center>Installing XFS</h1>
For a technical overview of XFS filesystem refer to <a
href="http://linux-xfs.sgi.com/projects/xfs/">SGI linux XFS page</a>
and to <a
href="http://linux-xfs.sgi.com/projects/xfs/publications.html">SGI
publications page</a>.<br>
Also see the <a href="http://oss.sgi.com/projects/xfs/faq.html">FAQ page</a>.
<p>
XFS is a journalling filesystem for Linux available from SGI.
It is a mature technology that has been proven on
IRIX systems as the default filesystem for all SGI customers.
XFS is licensed under GPL. <br>
XFS Linux 1.0 is released for the Linux 2.4 kernel, and I tried the
2.4.2 patch. So the first step is to acquire a linux-2.4.2 kernel from one
mirror of kernel.org.<br>
The patches are at <a
href="ftp://oss.sgi.com/projects/xfs/download/Release-1.0/patches/">oss.sgi.com/projects/xfs/download/Release-1.0/patches</a>.
From this directory download:
<ul>
<li>
<a
href="ftp://oss.sgi.com/projects/xfs/download/Release-1.0/patches/linux-2.4-xfs-1.0.patch.gz">linux-2.4-xfs-1.0.patch.gz</a>
<li>
<a
href="ftp://oss.sgi.com/projects/xfs/download/Release-1.0/patches/linux-2.4.2-core-xfs-1.0.patch.gz">linux-2.4.2-core-xfs-1.0.patch.gz</a>
<li>
<a
href="ftp://oss.sgi.com/projects/xfs/download/Release-1.0/patches/linux-2.4.2-kdb-04112001.patch.gz">linux-2.4.2-kdb-04112001.patch.gz</a>
</ul>
Copy the Linux kernel linux-2.4.2.tar.bz2 in <em>/usr/src</em> directory, rename the
existing linux directory to linux-old and extract the new kernel:
<pre>
mv linux linux-old
tar -Ixf inux-2.4.2.tar.bz2
</pre>
Copy each patch in the top directory of your linux source tree
(i.e. /usr/src/linux) and apply them:
<pre>
zcat patchfile.gz | patch -p1
</pre>
Then configure the kernel, enabling the options "XFS
filesystem support" (CONFIG_XFS_FS) and "Page Buffer support"
(CONFIG_PAGE_BUF) in the filesystem section.
Note that you will also need to upgrade the following system utilities to
these versions or later:
<ul>
<li><a
href="http://www.kernel.org/pub/linux/utils/kernel/modutils/v2.4/">modutils-2.4.0</a>
<li><a href="http://www.gnu.org/software/autoconf/autoconf.html">autoconf-2.13</a>
<li><a href="http://e2fsprogs.sourceforge.net/">e2fsprogs-devel-1.18</a>
</ul>
Install the new kernel and reboot.<br>
Now download the <a
href="ftp://oss.sgi.com/projects/xfs/download/Release-1.0/cmd_tars/xfsprogs-1.2.0.src.tar.gz">xfs
progs tools</a>. This tarball contains a set of commands to use the
XFS filesystem, such as mkfs.xfs. To build them:
<pre>
tar -zxf xfsprogs-1.2.0.src.tar.gz
cd xfsprogs-1.2.0
make configure
make
make install
</pre>
After installing this set of commands you can create a new XFS
filesystem with the command:
<pre>
mkfs -t xfs /dev/xxx
</pre>
One important option that you may need is "-f" which will force the
creation of a new filesystem, if a filesystem already exists on that
partition. Again, note that this will destroy all data currently on
that partition:
<pre>
mkfs -t xfs -f /dev/xxx
</pre>
You can then mount the new filesystem with the command:
<pre>
mount -t xfs /dev/xxx /mount_dir
</pre>
<h1 align=center>Installing ReiserFS</h1>
For technical notes about reiserFS refer to <a
href="http://www.namesys.com/">NAMESYS home page</a>
and to <a
href="http://www.namesys.com/faq.html">FAQ page</a>.<p>
ReiserFS has been in the official Linux kernel since
2.4.1-pre4. You always need to get the utils (e.g. mkreiserfs to
create ReiserFS on an empty partition, the resizer, etc.).<br>
The up-to-date ReiserFS version is available as a patch against either
2.2.x and 2.4.x kernels. I tested the patch against 2.2.19 Linux kernel.<p>
The first step, as usual, is to get a linux-2.2.19.tar.bz2 standard
kernel from a mirror of kernel.org.
Then get the reiserfs <a
href="ftp://ftp.namesys.com/pub/reiserfs-for-2.2/">2.2.19
patch</a>. At present time the last patch is 3.5.33.<br>
Please note that, if you choose to get the patch against 2.4.x kernel,
you should get also the utils tarball <a
href="ftp://ftp.namesys.com/pub/reiserfsprogs/reiserfsprogs-3.x.0j.tar.gz
">reiserfsprogs-3.x.0j.tar.gz</a>. <br>
Now unpack the kernel and the patch. Copy the tarballs in <em>/usr/src</em> and
move the linux directory to linux-old; then run the commands:
<pre>
tar -Ixf linux-2.2.19.tar.bz2
bzcat linux-2.2.19-reiserfs-3.5.33-patch.bz2 | patch -p0
</pre>
Compile the Linux kernel setting reiserfs support on filesystem section.<br>
Compile and install the reiserfs utils:
<pre>
cd /usr/src/linux/fs/reiserfs/utils
make
make install
</pre>
Install the new kernel and reboot.
Now you can create a new reiserfs filesystem with the command:
<pre>
mkreiserfs /dev/xxxx
</pre>
and mount it:
<pre>
mount -t reiserfs /dev/xxx /mount_dir
</pre>
<h1 align=center>Filesystems benchmark</h1>
For the test I used a Pentium III - 16 Mb RAM - 2 Gb HD with a Linux RedHat 6.2 installed. <br>
All the filesystems worked fine for me, so I started a little benchmark
analysis to compare their performances. As a first test I simulated a
crash turning off the power, in order to control the journal
recovery process. All filesystems passed successfully this phase and
the machine was on line in few
seconds with each filesystem. <p>
The next step is a benchmark analysis using bonnie++ program,
available at <a href="http://www.coker.com.au/bonnie++/">www.coker.com.au/bonnie++</a>.
The program tests database type access to a single file, and it tests
creation, reading, and
deleting of small files
which can simulate the usage of programs such as Squid, INN, or
Maildir-format programs (qmail). <br>
The benchmark command was:
<pre>
bonnie++ -d/work1 -s10 -r4 -u0
</pre>
which executes the test using 10Mb (-s10) in the
filesystem mounted in /work1 directory. So, before launching the
benchmark, you must create the requested filesystem on a partition
and mount it on /work1 directory. The other flags specify the RAM
amount in Mb (-r4) and the user (-u0, i.e. run as root).<p>
The results are shown in the following table.<p>
<table BORDER=3 CELLPADDING=2 NOSAVE >
<tr NOSAVE>
<td COLSPAN="2" NOSAVE class="header"></td>
<td COLSPAN="6" class="header"><b><font size=+2>Sequential Output</font></b></td>
<td COLSPAN="4" class="header"><b><font size=+2>Sequential Input</font></b></td>
<td COLSPAN="2" ROWSPAN="2" class="header"><b><font size=+2>Random</font></b>
<br><b><font size=+2>Seeks</font></b></td>
</tr>
<tr ALIGN=CENTER NOSAVE>
<td NOSAVE></td>
<td>Size:Chunk Size</td>
<td COLSPAN="2" BGCOLOR="#FFCC00" NOSAVE>Per Char</td>
<td COLSPAN="2" BGCOLOR="#FFFF00" NOSAVE>Block</td>
<td COLSPAN="2" BGCOLOR="#CCFFFF" NOSAVE>Rewrite</td>
<td COLSPAN="2" BGCOLOR="#FFCC00" NOSAVE>Per Char</td>
<td COLSPAN="2" BGCOLOR="#FFFF00" NOSAVE>Block</td>
</tr>
<tr>
<td COLSPAN="2"></td>
<td class="ksec"><font size=-2>K/sec</font></td>
<td class="ksec"><font size=-2>% CPU</font></td>
<td class="ksec"><font size=-2>K/sec</font></td>
<td class="ksec"><font size=-2>% CPU</font></td>
<td class="ksec"><font size=-2>K/sec</font></td>
<td class="ksec"><font size=-2>% CPU</font></td>
<td class="ksec"><font size=-2>K/sec</font></td>
<td class="ksec"><font size=-2>% CPU</font></td>
<td class="ksec"><font size=-2>K/sec</font></td>
<td class="ksec"><font size=-2>% CPU</font></td>
<td class="ksec"><font size=-2>/ sec</font></td>
<td class="ksec"><font size=-2>% CPU</font></td>
</tr>
<tr>
<td class="rowheader"><b><font size=+1>ext2</font></b></td>
<td BGCOLOR="#FFCCFF" NOSAVE class="size">10M</td>
<td>1471</td>
<td>97</td>
<td>14813</td>
<td>67</td>
<td>1309</td>
<td>14</td>
<td>1506</td>
<td>94</td>
<td>4889</td>
<td>15</td>
<td>309.8</td>
<td>10</td>
</tr>
<tr>
<td class="rowheader"></td>
<td class="size"></td>
<td></td>
</tr>
<tr>
<td class="rowheader"><b><font size=+1>ext3</font></b></td>
<td BGCOLOR="#FFCCFF" NOSAVE class="size">10M</td>
<td>1366</td>
<td>98</td>
<td>2361</td>
<td>38</td>
<td>1824</td>
<td>22</td>
<td>1482</td>
<td>94</td>
<td>4935</td>
<td>14</td>
<td>317.8</td>
<td>10</td>
</tr>
<tr>
<td class="rowheader"></td>
<td class="size"></td>
<td></td>
</tr>
<tr>
<td class="rowheader"><b><font size=+1>xfs</font></b></td>
<td BGCOLOR="#FFCCFF" NOSAVE class="size">10M</td>
<td>1206</td>
<td>94</td>
<td>9512</td>
<td>77</td>
<td>1351</td>
<td>33</td>
<td>1299</td>
<td>98</td>
<td>4779</td>
<td>80</td>
<td>229.1</td>
<td>11</td>
</tr>
<tr>
<td class="rowheader"></td>
<td class="size"></td>
<td></td>
</tr>
<tr>
<td class="rowheader"><b><font size=+1>reiserfs</font></b></td>
<td BGCOLOR="#FFCCFF" NOSAVE class="size">10M</td>
<td>1455</td>
<td>99</td>
<td>4253</td>
<td>31</td>
<td>2340</td>
<td>26</td>
<td>1477</td>
<td>93</td>
<td>5593</td>
<td>26</td>
<td>174.3</td>
<td>5</td>
</tr>
</table>
<p>
<table BORDER=3 CELLPADDING=2 NOSAVE >
<tr NOSAVE>
<td COLSPAN="2" class="header"></td>
<td COLSPAN="6" class="header"><b><font size=+2>Sequential Create</font></b></td>
<td COLSPAN="6" class="header"><b><font size=+2>Random Create</font></b></td>
</tr>
<tr>
<td> </td>
<td>Num Files</td>
<td COLSPAN="2" BGCOLOR="#66FF99" NOSAVE>Create</td>
<td COLSPAN="2" BGCOLOR="#33FF33" NOSAVE>Read</td>
<td COLSPAN="2" BGCOLOR="#33CC00" NOSAVE>Delete</td>
<td COLSPAN="2" BGCOLOR="#66FF99" NOSAVE>Create</td>
<td COLSPAN="2" BGCOLOR="#33FF33" NOSAVE>Read</td>
<td COLSPAN="2" BGCOLOR="#33CC00" NOSAVE>Delete</td>
</tr>
<tr>
<td></td>
<td> </td>
<td class="ksec"><font size=-2>/ sec</font></td>
<td class="ksec"><font size=-2>% CPU</font></td>
<td class="ksec"><font size=-2>/ sec</font></td>
<td class="ksec"><font size=-2>% CPU</font></td>
<td class="ksec"><font size=-2>/ sec</font></td>
<td class="ksec"><font size=-2>% CPU</font></td>
<td class="ksec"><font size=-2>/ sec</font></td>
<td class="ksec"><font size=-2>% CPU</font></td>
<td class="ksec"><font size=-2>/ sec</font></td>
<td class="ksec"><font size=-2>% CPU</font></td>
<td class="ksec"><font size=-2>/ sec</font></td>
<td class="ksec"><font size=-2>% CPU</font></td>
</tr>
<tr>
<td class="rowheader"><b><font size=+1>ext2</font></b></td>
<td BGCOLOR="#FFCCFF" NOSAVE>16</td>
<td>94</td>
<td>99</td>
<td>278</td>
<td>99</td>
<td>492</td>
<td>97</td>
<td>95</td>
<td>99</td>
<td>284</td>
<td>100</td>
<td>93</td>
<td>41</td>
</tr>
<tr>
<td class="rowheader"><b><font size=+1>ext3</font></b></td>
<td BGCOLOR="#FFCCFF" NOSAVE>16</td>
<td>89</td>
<td>98</td>
<td>274</td>
<td>100</td>
<td>458</td>
<td>96</td>
<td>93</td>
<td>99</td>
<td>288</td>
<td>99</td>
<td>97</td>
<td>45</td>
</tr>
<tr>
<td class="rowheader"><b><font size=+1>xfs</font></b></td>
<td BGCOLOR="#FFCCFF" NOSAVE >16</td>
<td>92</td>
<td>99</td>
<td>251</td>
<td>96</td>
<td>436</td>
<td>98</td>
<td>91</td>
<td>99</td>
<td>311</td>
<td>99</td>
<td>90</td>
<td>41</td>
</tr>
<tr>
<td class="rowheader"><b><font size=+1>reiserfs</font></b></td>
<td BGCOLOR="#FFCCFF" NOSAVE>16</td>
<td>1307</td>
<td>100</td>
<td>8963</td>
<td>100</td>
<td>1914</td>
<td>99</td>
<td>1245</td>
<td>99</td>
<td>9316</td>
<td>100</td>
<td>1725</td>
<td>100</td>
</tr>
</table>
<p>
Two data are shown for each test: the speed of the filesystem (in
K/sec) and the CPU usage (in %). The higher the speed the better the
filesystem. The opposite is true for the CPU usage. <br>
As you can see reiserFS reports a hands down victory in managing files
(section <em>Sequential Create</em> and <em>Random Create</em>),
overwhelming its opponents by a factor higher than 10. In addition to
that is almost as good as the other filesystem in the <em>Sequential
Output</em> and <em>Sequential Input</em>.
There isn't any significant difference among the other filesystems. XFS
speed is similar to ext2 filesystem, and ext3 is, as expected, a
little slower than ext2 (it is basically the same thing, and it wastes
some time during the journalling calls). <p>
As a last test I get the mongo benchmark program available at reiserFS
benchmark page at
<a href="http://www.namesys.com">www.namesys.com</a>,
and I modified it in order to test the three journalling filesystems. I
inserted in the mongo.pl perl script the commands to mount the xfs and
ext3 filesystem and to format them. Then I started a benchmark analysis.<br>
The script formats partition /dev/xxxx, mounts it and runs given number
of processes during each phase: Create, Copy, Symlinks, Read, Stats,
Rename and Delete. Also, the program calculates fragmentation after Create
and Copy phases:
<pre>
Fragm = number_of_fragments / number_of_files
</pre>
You can find the same results in the directory <em>results</em> in the files:
<pre>
log - raw results
log.tbl - results for compare program
log_table - results in table form
</pre>
The tests was executed as in the following example:
<pre>
mongo.pl ext3 /dev/hda3 /work1 logext3 1
</pre>
where ext3 must be replaced by reiserfs or xfs in order to test the
other filesystems. The other arguments are the device to mount, where
the filesystem to test is located, the mounting directory, the
filename where the results are stored and the number of processes to
start.<p>
In the following tables there are the results of this analysis. The data
reported is time (in sec). The lower the value, the better the
filesystem.
In the first table the median dimension of files managed is 100 bytes, in the
second one it is 1000 bytes and in the last one 10000 bytes.<p>
<table BORDER NOSAVE >
<tr ALIGN=LEFT VALIGN=TOP NOSAVE>
<td NOSAVE><tt></tt>
<br><font size=-1></font></td>
<td BGCOLOR="#FFCCFF" NOSAVE><tt>ext3</tt>
<br><tt><font size=-1>files=68952</font></tt>
<br><tt><font size=-1>size=100 bytes</font></tt>
<br><tt><font size=-1>dirs=242</font></tt></td>
<td BGCOLOR="#FFCCFF" NOSAVE><tt>XFS</tt>
<br><tt><font size=-1>files=68952</font></tt>
<br><tt><font size=-1>size=100 bytes</font></tt>
<br><tt><font size=-1>dirs=241</font></tt></td>
<td BGCOLOR="#FFCCFF" NOSAVE><tt>reiserFS</tt>
<br><tt><font size=-1>files=68952</font></tt>
<br><tt><font size=-1>size=100 bytes</font></tt>
<br><tt><font size=-1>dirs=241</font></tt></td>
</tr>
<tr>
<td><tt>Create</tt></td>
<td>90.07</td>
<td>267.86</td>
<td>53.05</td>
</tr>
<tr>
<td><tt>Fragm.</tt></td>
<td>1.32</td>
<td>1.02</td>
<td>1.00</td>
</tr>
<tr>
<td><tt>Copy</tt></td>
<td>239.02</td>
<td>744.51</td>
<td>126.97</td>
</tr>
<tr>
<td><tt>Fragm.</tt></td>
<td>1.32</td>
<td>1.03</td>
<td>1.80</td>
</tr>
<tr>
<td><tt>Slinks</tt></td>
<td>0</td>
<td>203.54</td>
<td>105.71</td>
</tr>
<tr>
<td><tt>Read</tt></td>
<td>782.75</td>
<td>1543.93</td>
<td>562.53</td>
</tr>
<tr>
<td><tt>Stats</tt></td>
<td>108.65</td>
<td>262.25</td>
<td>225.32</td>
</tr>
<tr>
<td><tt>Rename</tt></td>
<td>67.26</td>
<td>205.18</td>
<td>70.72</td>
</tr>
<tr>
<td><tt>Delete</tt></td>
<td>23.80</td>
<td>389.79</td>
<td>85.51</td>
</tr>
</table>
<p>
<table BORDER NOSAVE >
<tr ALIGN=LEFT VALIGN=TOP NOSAVE>
<td NOSAVE><tt></tt>
<br></td>
<td BGCOLOR="#FFCCFF" NOSAVE><tt>ext3</tt>
<br><tt><font size=-1>files=11248</font></tt>
<br><tt><font size=-1>size=1000 bytes</font></tt>
<br><tt><font size=-1>dirs=44</font></tt></td>
<td BGCOLOR="#FFCCFF" NOSAVE><tt>XFS</tt>
<br><tt><font size=-1>files=11616</font></tt>
<br><tt><font size=-1>size=1000 bytes</font></tt>
<br><tt><font size=-1>dirs=43</font></tt></td>
<td BGCOLOR="#FFCCFF" NOSAVE><tt>ReiserFS</tt>
<br><tt>f<font size=-1>iles=11616</font></tt>
<br><tt><font size=-1>size=1000 bytes</font></tt>
<br><tt><font size=-1>dirs=43</font></tt></td>
</tr>
<tr>
<td><tt>Create</tt></td>
<td>30.68</td>
<td>57.94</td>
<td>36.38</td>
</tr>
<tr>
<td><tt>Fragm.</tt></td>
<td>1.38</td>
<td>1.01</td>
<td>1.03</td>
</tr>
<tr>
<td><tt>Copy</tt></td>
<td>75.21</td>
<td>149.49</td>
<td>84.02</td>
</tr>
<tr>
<td><tt>Fragm.</tt></td>
<td>1.38</td>
<td>1.01</td>
<td>1.43</td>
</tr>
<tr>
<td><tt>Slinks</tt></td>
<td>16.68</td>
<td>29.59</td>
<td>19.29</td>
</tr>
<tr>
<td><tt>Read</tt></td>
<td>225.74</td>
<td>348.99</td>
<td>409.45</td>
</tr>
<tr>
<td><tt>Stats</tt></td>
<td>25.60</td>
<td>46.41</td>
<td>89.23</td>
</tr>
<tr>
<td><tt>Rename</tt></td>
<td>16.11</td>
<td>33.57</td>
<td>20.69</td>
</tr>
<tr>
<td><tt>Delete</tt></td>
<td>6.04</td>
<td>64.90</td>
<td>18.21</td>
</tr>
</table>
<p>
<table BORDER NOSAVE >
<tr ALIGN=LEFT VALIGN=TOP NOSAVE>
<td NOSAVE><tt></tt>
<br></td>
<td BGCOLOR="#FFCCFF" NOSAVE><tt>ext3</tt>
<br><tt><font size=-1>files=2274</font></tt>
<br><tt><font size=-1>size=10000 bytes</font></tt>
<br><tt><font size=-1>dirs=32</font></tt></td>
<td BGCOLOR="#FFCCFF" NOSAVE><tt>XFS</tt>
<br><tt><font size=-1>files=2292</font></tt>
<br><tt><font size=-1>size=10000 bytes</font></tt>
<br><tt><font size=-1>dirs=31</font></tt></td>
<td BGCOLOR="#FFCCFF" NOSAVE><tt>reiserFS</tt>
<br><tt><font size=-1>files=2292</font></tt>
<br><tt><font size=-1>size=10000 bytes</font></tt>
<br><tt><font size=-1>dirs=31</font></tt></td>
</tr>
<tr>
<td><tt>Create</tt></td>
<td>27.13</td>
<td>25.99</td>
<td>22.27</td>
</tr>
<tr>
<td><tt>Fragm.</tt></td>
<td>1.44</td>
<td>1.02</td>
<td>1.05</td>
</tr>
<tr>
<td><tt>Copy</tt></td>
<td>55.27</td>
<td>55.73</td>
<td>43.24</td>
</tr>
<tr>
<td><tt>Fragm.</tt></td>
<td>1.44</td>
<td>1.02</td>
<td>1.12</td>
</tr>
<tr>
<td><tt>Slinks</tt></td>
<td>1.33</td>
<td>2.51</td>
<td>1.43</td>
</tr>
<tr>
<td><tt>Read</tt></td>
<td>40.51</td>
<td>50.20</td>
<td>56.34</td>
</tr>
<tr>
<td><tt>Stats</tt></td>
<td>2.34</td>
<td>1.99</td>
<td>3.52</td>
</tr>
<tr>
<td><tt>Rename</tt></td>
<td>0.99</td>
<td>1.10</td>
<td>1.25</td>
</tr>
<tr>
<td><tt>Delete</tt></td>
<td>3.40</td>
<td>8.99</td>
<td>1.84</td>
</tr>
</table>
<p>
From these tables you can see that ext3 is usually faster in Stats
Delate and Rename, while reiserFS wins in Create and Copy. Also note
that the performance of reiserFS in better in the first case (small
files) as expected by its technical documentation.
<h1 align=center>Conclusions</h1>
There are at present time at least two robust and reliable journalling
filesystems for Linux (i.e. XFS and reiserFS) which can be utilized
without fear.<br>
ext3 is still an alpha release and can undergo several failures. I had
some
problems using bonnie++ on this filesystem: the system reported some
VM
errors and killed the shell I was using. <p>
Considering the benchmark results my advice is to install a reiserFS
filesystem in the future (I'll surely do it).
<!-- *** BEGIN bio *** -->
<SPACER TYPE="vertical" SIZE="30">
<P>
<H4><IMG ALIGN=BOTTOM ALT="" SRC="../gx/note.gif">Matteo Dell'Omodarme</H4>
<CITE>I'm a student at the University of Pisa and a Linux user since 1994.
Now I'm working on the administrations of Linux boxes at the Astronomy section
of the Department of Physics, with special experience about security. My
primary email address is
<A HREF="mailto:matt@martine2.difi.unipi.it">matt@martine2.difi.unipi.it</A>.
</CITE>
<!-- *** END bio *** -->
<!-- *** BEGIN copyright *** -->
<P> <hr> <!-- P -->
<H5 ALIGN=center>
Copyright &copy; 2001, Matteo Dell'Omodarme.<BR>
Copying license <A HREF="../copying.html">http://www.linuxgazette.com/copying.html</A><BR>
Published in Issue 68 of <i>Linux Gazette</i>, July 2001</H5>
<!-- *** END copyright *** -->
<!--startcut ==========================================================-->
<HR><P>
<CENTER>
<!-- *** BEGIN navbar *** -->
<IMG ALT="" SRC="../gx/navbar/left.jpg" WIDTH="14" HEIGHT="45" BORDER="0" ALIGN="bottom"><A HREF="collinge.html"><IMG ALT="[ Prev ]" SRC="../gx/navbar/prev.jpg" WIDTH="16" HEIGHT="45" BORDER="0" ALIGN="bottom"></A><A HREF="index.html"><IMG ALT="[ Table of Contents ]" SRC="../gx/navbar/toc.jpg" WIDTH="220" HEIGHT="45" BORDER="0" ALIGN="bottom" ></A><A HREF="../index.html"><IMG ALT="[ Front Page ]" SRC="../gx/navbar/frontpage.jpg" WIDTH="137" HEIGHT="45" BORDER="0" ALIGN="bottom"></A><A HREF="http://www.linuxgazette.com/cgi-bin/talkback/all.py?site=LG&article=http://www.linuxgazette.com/issue68/dellomodarme.html"><IMG ALT="[ Talkback ]" SRC="../gx/navbar/talkback.jpg" WIDTH="121" HEIGHT="45" BORDER="0" ALIGN="bottom" ></A><A HREF="../faq/index.html"><IMG ALT="[ FAQ ]" SRC="./../gx/navbar/faq.jpg"WIDTH="62" HEIGHT="45" BORDER="0" ALIGN="bottom"></A><A HREF="ghosh.html"><IMG ALT="[ Next ]" SRC="../gx/navbar/next.jpg" WIDTH="15" HEIGHT="45" BORDER="0" ALIGN="bottom" ></A><IMG ALT="" SRC="../gx/navbar/right.jpg" WIDTH="15" HEIGHT="45" ALIGN="bottom">
<!-- *** END navbar *** -->
</CENTER>
</BODY></HTML>
<!--endcut ============================================================-->
Reference in New Issue View Git Blame Copy Permalink