All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
1. Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
3. Neither the name of the University nor the names of its contributors
may be used to endorse or promote products derived from this software
without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
SUCH DAMAGE.
@(#)7.t 5.1 (Berkeley) 4/17/91
Conclusions
4.2BSD, while functionally superior to 4.1BSD, lacked much of the performance tuning required of a good system. We found that the distributed system spent 10-20% more time in the kernel than 4.1BSD. This added overhead combined with problems with several user programs severely limited the overall performance of the system in a general timesharing environment.
Changes made to the system since the 4.2BSD distribution have eliminated most of the added system overhead by replacing old algorithms or introducing additional cacheing schemes. The combined caches added to the name translation process reduce the average cost of translating a pathname to an inode by more than 50%. These changes reduce the percentage of time spent running in the system by nearly 9%.
The use of silo input on terminal ports only when necessary has allowed the system to avoid a large amount of software interrupt processing. Observations show that the system is forced to field about 25% fewer interrupts than before.
The kernel changes, combined with many bug fixes, make the system much more responsive in a general timesharing environment. The 4.3BSD Berkeley UNIX system now appears capable of supporting loads at least as large as those supported under 4.1BSD while providing all the new interprocess communication, networking, and file system facilities. .nr H2 1
\s+2Acknowledgements\s0We would like to thank Robert Elz for sharing his ideas and his code for cacheing system wide names and searching the process table. We thank Alan Smith for initially suggesting the use of a capability based cache. We also acknowledge George Goble who dropped many of our changes into his production system and reported back fixes to the disasters that they caused. The buffer cache read-ahead trace package was based on a program written by Jim Lawson. Ralph Campbell implemented several of the C library changes. The original version of the Internet daemon was written by Bill Joy. In addition, we would like to thank the many other people that contributed ideas, information, and work while the system was undergoing change. .nr H2 1
\s+2References\s-2