os/doc
3
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

- Add 2013Q4 report on cam(4) and geom(4) improvements

Browse source 
Submitted by:	mav
This commit is contained in:
Gabor Pali 2014-01-11 10:29:52 +00:00
parent b981f11356
commit a750aeeb18
Notes: svn2git 2020-12-08 03:00:23 +00:00 
svn path=/head/; revision=43470
1 changed files with 50 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

51
en_US.ISO8859-1/htdocs/news/status/report-2013-10-2013-12.xml
View file

@ -19,7 +19,7 @@
<!-- XXX: Keep updating the number of entries -->
<p>Thanks to all the reporters for the excellent work! This report
contains 10 entries and we hope you enjoy reading it.</p>
contains 11 entries and we hope you enjoy reading it.</p>
<p>The deadline for submissions covering between January and
March 2014 is April 7th, 2014.</p>
@ -559,4 +559,53 @@
further.</task>
</help>
</project>
<project cat='kern'>
<title>GEOM Direct Dispatch and Fine-Grained CAM Locking</title>
<contact>
<person>
<name>
<given>Alexander</given>
<common>Motin</common>
</name>
<email>mav@FreeBSD.org</email>
</person>
</contact>
<links>
<url href="http://people.freebsd.org/~mav/disk.pdf">Slides from EuroBSDCon 2013, also describing this project</url>
<url href="http://svnweb.freebsd.org/changeset/base/260387">CAM improvements in the stable/10 branch</url>
<url href="http://svnweb.freebsd.org/changeset/base/260385">GEOM improvements in the stable/10 branch</url>
</links>
<body>
<p>The CAM and GEOM multi-processor scalability improvement
project has completed. The corresponding code has been committed
to &os; <tt>head</tt> and recently merged to the
<tt>stable/10</tt> branch, it shall appear in
<tt>10.1-RELEASE</tt>.</p>
<p>As part of this project, <tt>cam(4)</tt> (ATA/SCSI subsystem)
has received more fine-grained locking for better utilization of
multi-core systems. In addition, locking in <tt>geom(4)</tt>
(block storage subsystem) has been also polished, and a new
direct dispatch functionality was implemented to spread the load
between multiple threads and processors, and reduce the number
of context switches.</p>
<p>Thanks to these <tt>cam(4)</tt> and <tt>geom(4)</tt> changes,
the peak I/O rate has doubled on comptemporary hardware, reaching
up to 1,000,000&nbsp;IOPS!</p>
<p>The project was sponsored by iXsystems, Inc.</p>
</body>
<help>
<task>Some CAM controller drivers (SIMs) could be also optimized
to get more benefits from this project, utilizing the new locking
models and direct command completions from multiple interrupt
threads.</task>
</help>
</project>
</report>