![]() |
![]() |
![]() |
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
![]() |
![]() |
To access the contents, click the chapter and section titles.
Oracle Performance Tuning and Optimization
CPU EnhancementsEnhancing the CPUs on your SMP or MPP system can provide instantaneous performance improvements (assuming that you are not I/O bound). The speed of CPUs is constantly being improved, as are new and better cache designs. Although the backup process primarily focuses on reading the disk and writing to another device, many backup programs include efficient compression options. Because you usually are limited by the speed of your tape drive and the speed of the network for network backups, compressing the data allows you to increase the performance of the backup by reducing the amount of data that must be transferred. If you have sufficiently fast CPUs, you can perform efficient compression. If you add CPUs to your system, you can reduce the effect the backup or compression has on other users in the system. Make sure that the speed of the CPUs is sufficiently fast to compress the backup data at a rate the meets or exceeds the performance of the slowest component in the backup process. Doing so guarantees that the compression process is not a bottleneck in backup process. The way to test this is simple: Back up a typical table twice: once using compression and again without using compression; then compare the results. I/O EnhancementsBy enhancing the I/O devices to be optimized, you can see significant performance increases during backup. Disk arrays allow you to perform a hot backup much more effectively than with traditional Oracle striping techniques. The use of a disk array can be enhanced by using a sufficiently large block size to cause multiple disks to be active simultaneously. Another benefit of hardware disk arrays is caching. Most disk arrays on the market today offer some type of write or read/write cache on the controller. The effect of this cache is to improve the speed of writing to the disk as well as to mask the overhead associated with fault tolerance. Backup activities may see improved performance with disk controllers that take advantage of read-ahead features. Read-ahead occurs when the controller detects a sequential access, reads an entire track or some other large amount of data, and caches the additional data in anticipation of a request from the OS. Because this data can then be accessed almost immediately, having data in the controller cache can be a benefit. Enhancements to the I/O subsystem almost always help backups because large amounts of data are read during backups. Be sure that you have a high-speed I/O path, properly configured. An I/O bottleneck is usually difficult to work around. A good third-party backup program can also be beneficial. Network EnhancementsIf you perform backups over a network, the speed of the network is extremely important, as is the amount of other traffic on the network. By segmenting the network into a production OLTP network and a backup network, you can see a reduced effect on the OLTP users and an increase in backup performance. This network segment should be of a high-performance variety, such as 100Base-T or fiber optics. The less time you spend waiting for the network, the less time the backup takes. Because it is easy to improve network performance by adding additional links or faster links, it is a good investment to try this. Split Up the BackupOne way you can minimize the time the systems performance is reduced because of backups is to split the backup into several portions. The backup can be done on a tablespace basis: you can back up some tablespaces on one day and back up other tablespaces on other days. Doing so can reduce the amount of time required to perform the backup on a daily basis. However, the total time necessary to perform the backup remains the same. When segmenting backups, you can decide which tablespaces have more critical data or are updated most frequently and back up those tablespaces more frequently than other, less-used tablespaces. In this manner, you can reduce the time it takes to restore your most critical data. In the event of a system failure, remember that the time it takes to restore your data depends on when the last backup occurred and how much data has been modified or added to that tablespace since that backup. By performing backups more frequently on active tables, you can reduce the time it takes to restore the system. A reasonable schedule would allow you to perform cold backups during non-peak hours. You can split up the backup into several sessions. By doing this, you can perform a backup over several days, reducing the time each day that the systems performance is affected. If backup takes several days, you may want to back up the most active, critical data more often. Here are some recommendations for segmenting backups:
|
![]() |
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. |