Click Here!
home account info subscribe login search My ITKnowledge FAQ/help site map contact us


 
Brief Full
 Advanced
      Search
 Search Tips
To access the contents, click the chapter and section titles.

Oracle Performance Tuning and Optimization
(Publisher: Macmillan Computer Publishing)
Author(s): Edward Whalen
ISBN: 067230886x
Publication Date: 04/01/96

Bookmark It

Search this book:
 
Previous Table of Contents Next


Server OS Tuning

You may have to tune the server OS to facilitate the large number of processes and network connections required of an OLTP system. The server OS also affects the optimization of I/O performance. Following are some of the things you may have to tune in the server OS; keep in mind that some OSes may not require any tuning in these areas:

  Processes. The system must be tuned to handle large numbers of processes.
  Memory. The system should be tuned to reduce unnecessary memory usage so that Oracle can use as much of the system’s memory as possible for the SGA and server processes.
  Memory enhancements. Take advantage of 4M pages and ISM if they are available. Both of these features improve Oracle performance in an OLTP environment.
  I/O. If necessary, tune I/O to allow for optimal performance and use of AIO.
  Scheduler. If possible, turn off preemptive scheduling and load balancing. In an OLTP environment, these features simply waste CPU cycles.
  Cache affinity. If possible, turn off cache affinity. In a heavily used OLTP environment, cache affinity does not benefit you; it only wastes CPU cycles.
  Network. Tune the system to favor your network protocol. Also configure the network for optimal Oracle performance. You may want to change the network packet size.

The server operating system is mainly a host on which Oracle does its job. Any work done by the operating system is essentially overhead for Oracle. By optimizing code paths and reducing OS overhead, you can increase Oracle performance. This topic is discussed in more detail in Chapter 12, “Operating System-Specific Tuning.”

Enhancements

You may want to make some additional enhancements to improve the performance of the OLTP system. Listed here are some of the items that can offer performance improvement; also listed are a few items that do not improve performance. Each item is accompanied by an explanation.

  Block size. Because typical OLTP transactions are small in nature and many of them happen simultaneously, it is not beneficial to increase the database block size. The default value of 2K is probably sufficient.
  Clusters. Depending on the application, you may benefit from clusters. The benefit you receive depends exclusively on the data access patterns and your application.
  Hash clusters. OLTP systems typically access tables with some sort of ID, such as a part number or account ID. If this is the case in your application, you may enhance performance by creating hash clusters on some of those tables.
  Indexes. Based on how the data is accessed, you may see significant performance benefits by using indexes. How and when to use indexes is described in Chapter 9, “Oracle Instance Tuning.”
  Multiblock reads and writes. In the OLTP environment, it is very unlikely that you will benefit from multiblock reads and writes because very little data access involves sequential access to data.
  Parallel query. The Oracle Parallel Query option does not offer any benefit in the OLTP environment. Parallelizing queries offers the benefit of making long-running queries concurrent. Because most OLTP transactions are small and of short duration, there is not much of a benefit.

Many of these enhancements can help your performance. The specific effect on your system varies, but the information presented here is true for most systems.

Oracle Parallel Server Option

The Oracle Parallel Server option may be very beneficial to your operation in two ways:

1.  Performance. If your system is a candidate for the Parallel Server option, you should see significant performance improvements.
2.  Fault tolerance. Many OLTP installations demand no downtime. By using the Oracle Parallel Server option, you can keep the system running, even if one computer fails.

The Oracle Parallel Server option can enhance the performance of your OLTP system—but only if the system is suitable for the Parallel Server option. To see scaleable performance increases, the system must meet the following criteria:

  Partitionable data. The data should be partitionable. In other words, the work related to certain tables should be done on one computer while work related to other tables should be done on another computer. If all users access the exact same data, your application may not be a good candidate for the Parallel Server option. Although some overlap is fine, the majority of the data should be partitionable.
  Many processes. If you have only a few users and concurrent transactions, your system is probably not a good candidate for the Parallel Server option.

If these criteria are met, you can probably benefit from using the Oracle Parallel Server option.


NOTE:  The Oracle Parallel Server option is not available on all platforms. Consult your system provider or Oracle to determine which platforms support the Oracle Parallel Server option.


Previous Table of Contents Next


Products |  Contact Us |  About Us |  Privacy  |  Ad Info  |  Home

Use of this site is subject to certain Terms & Conditions, Copyright © 1996-2000 EarthWeb Inc.
All rights reserved. Reproduction whole or in part in any form or medium without express written permission of EarthWeb is prohibited.