![]() |
![]() |
![]() |
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
![]() |
![]() |
To access the contents, click the chapter and section titles.
Oracle Performance Tuning and Optimization
Analysis During the analysis phase, you should look at two things: First, is the custom benchmark performing as specified? Does it generate the workload you want or is there something wrong? Second, how does the system perform? If the benchmark contains a design flaw or a coding bug, it is essential to determine this as soon as possible rather than proceeding with decisions based on bad data. Verify that the results look like what you expect them to; if they dont, ask the question why? Once you verify that the test is working properly, it is time to analyze the results. If you are running the test to determine which competing product to buy or to determine whether an application change has an effect on the system, it is important to retune the system after any changes are made. Running the same test on two different hardware platforms may give deceiving results if you dont analyze the data correctly. Heres an example: By using the same tuning parameters on System A and System B, you may see better results on System A because of an architectural feature in the hardware. By tuning both systems individually to their highest potential, you may find that System B performs much better. Each system is different; it is invalid to hold tuning parameters constant for comparisons. The only way to effectively compare systems is to tune each as well as you can. SummaryThis chapter looked at the industry standard benchmarks from the TPC, briefly described the publication benchmarks, and outlined how to design and implement custom benchmarks. Both standard and custom benchmarks serve a dual purpose: an indicator of performance and a verification tool when changes have been made. You should now have an idea about the effort required to produce a TPC benchmark as well as the effort necessary to implement your own benchmark. These tools can be very powerful and useful once they have been implemented. This chapter also described the various TPC benchmarks available and how to interpret those results. The TPC benchmark results are available by accessing the TPC Web site and can be a powerful tool in helping you decide which platform is right for you. Because TPC benchmarks are published by the vendors who have a lot at stake, you can be assured that the configurations are optimized and well tuned. Similarly, for custom benchmarks, you should ensure that each system you are comparing is equally well tuned.
|
![]() |
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. |