rather than a separate test MySQL main reasons are the following

[hvkyez49]

2 benchmark strategy
There are two main benchmark strategies: First, the overall test for the entire system, in addition to a separate test MySQL. These two strategies is also known as integrated (fullstack) and single-component formula (singlecomponent) benchmark for the entire system as an integrated test,Australia Ugg Boots, rather than a separate test MySQL main reasons are the following:
Test the entire application, including Web servers y, application code, networks and databases are very useful because the user not only concern the performance of MySQL itself, but the overall performance of the application is not always the bottleneck MySQL, through the whole The test can reveal that the overall test is only done on the application, can be found between the various parts of the overall impact of the application cache integrated application testing can better reveal the true performance, and individual components of the test difficult to do this On the other hand, the application of the overall benchmark is difficult to establish, or even to set up correctly. If the benchmark design problems, the results will not reflect the real situation,ugg outlet uk, and thus make decisions based on this it may be wrong, however, sometimes do not understand the whole application,ugg boots sale, but only need to focus on the performance of MySQL,http://www.wsmithauction.com/ugg-canada-sale.html, at least early in the project can do so. Based on the following situation,http://www.bc4x4.com/ugg-boots-canada.asp, you can choose to only test MySQL:
To compare different properties schema or query a particular problem for the application of the test in order to avoid lengthy benchmark through a short benchmark, do a quick benchmark test is also useful, but the data itself and the data set size should be true. If possible, use a snapshot of data production Unfortunately, real data set a benchmark based complex and time-consuming. If you can get a copy of the production data sets, of course, very lucky, but this is usually not possible. For example, to test a newly developed new applications, it is only a few users and data. If you want to test the application on a large scale expansion to future performance, it is only by simulating a large number of data and pressure to perform.
相关的主题文章: