Hammerdb database pressure tools using the brief steps

Welcome reprint, Reprint please indicate source: http://blog.csdn.net/notbaron/article/details/38879681

Hammerdb database pressure tools using the brief steps

Although there is no diagram, but the text description is as in place.

Can be described by the language steps, the tool is very simple to use, can simulate the TPC-C test model, the unit of test results is TPM, note is not tpmc.

The TPM represents the number of transaction transactions per minute. TPMC is the transaction unit of TPC-C.

The overall steps are as follows:

1. Installed operating system of the machine under test

2. If ASM is used, install the grid and skip this step if you do not use ASM.

3, install the database software, just software. Specific steps the toad skipped, you can search for the database version of the Web. Mainly environment variables, RPM package requirements.

4, Oracle user, using the DBCA command to build the library. Both the DB instance name and Global name are set to hammer, then the SYS and system passwords are changed to a simple one, such as Oracle

5, after the database is created, use NETCA to configure network monitoring. The listener name is set to hammer.

6. Create a table space, you can create one or you can create two table spaces. (This test must have user tablespace, test it) here Toad creates two one is the tablespace name is TPCC and the other is tpcctab. A size of 100G is sufficient (for size and test database size)

7, install HAMMERDB package (the official website can be downloaded, directly installed, the need for some RPM package, through the Yum installation. ) (in the machine to be tested or alone to get a machine when the client can also)

But getting a client alone is putting pressure on the machine under test. The network becomes a bottleneck.

7, open the Hammerdb interface, the first step is the CREATE schema interface (there is an option to remember). Set the necessary parameters such as SID, database tablespace name for test, System password, number of warehouse, number of users, etc.

When finished, double-click Create to import data (depending on the number of warehouse set, the storage media performance determines the import data time).

8, set the test script, there is an options option, set the number of users and so on. Then double-clicking Load,hammerdb will load the number of users.

9, set the number of test users, there is an options option, after setting up, double-click Create, and then click the Green Go button on the image interface to test.

10, before the 9th step, it is better to set the transaction inside the parameters, so that you can visualize the performance of the result values.

Finally, say something about Oracle tuning:

1, Memory large page open

2, the Pfile parameter setting of the database

3, the main database table space storage location (storage media directly affect database performance), redo and data are placed separately.

Data in the random read performance of the media, such as SSD Magnetic array, SSD card, Fusionio, etc., the smaller the delay the better. Redo data is placed on sequential media to ensure bandwidth size. If the redo has a bottleneck,

The data storage media is no longer a plus.

4, if there is a network, set up network interrupt binding

Bottleneck Analysis:

1, the system level analysis, Cpu,mem,io a traverse check to ensure that the hardware does not appear bottlenecks. If this problem occurs in advance, change the hardware.

2, if the hardware OK, and CPU utilization is not up, then the output AWR report analysis, this analysis requires relatively professional knowledge, slowly accumulate it.

If these two points are done thoroughly, well, that's basically everything, but the fact is that either point requires professional analytical tools and analytical level.

OK, just write it down. Begins as an Oracle database.