Performance testing of Oracle databases using Swingbench __python

1,swingben Introduction

The current open source Oracle Stress testing Tools on the Web are mainly ORABM and swingbench, and because ORABM does not support Oracle 11g versions, the test uses Swingben for stress testing. In addition, Swingbench can also test the RAC. Swingbench is an Oracle stress testing tool developed by the UK based Oracle Database Solutions Group, as if it were an officially defunct project, the official page http://dominicgiles.com/    The latest version of the software can be downloaded on swingbench.html. Swingbench can be run on Windows and Linux platforms, this test uses Linux platform, the specific process is as follows
1, using Oewizard to import test data, you can follow the procedure prompts for data import. The connection string is IP:port:SID and the selection type is OCI
2, set the environment variable Java version requires more than 1.6, the connection string for 10.4.x.x:1521:x
Export path=/monitor/agent_12c/core/12.1.0.1.0/jdk/bin/: $PATH
Export display=10.6.*.*:0.0
cd/monitor/swingbench/swingbench/bin/
3, use the Swingbench homepage to carry on the pressure test

2, Pressure test results

The server type is an IBM X3755M3 server configured with 4CPU *8core,128g memory. The test results showed that the peak reached 480,000 Tpmc,tps the largest close to 10,000. At this point, database host CPU utilization is close to 95%

Top-15:26:34 up 6 days, 21:57, 2 users, Load average:189.06, 121.48, 58.92

tasks:909 Total, running, 873 sleeping, 0 stopped, 0 zombie

Cpu (s): 79.1%us, 12.9%sy, 0.0%ni, 6.5%id, 0.2%wa, 0.1%hi, 1.2%si, 0.0%st

Appendix:

Related test parameters for A,swingbench

1. Users:the number of users on Swingbench GUI (threads) that attach to a database and the amount and tye of work they perfor M. users can dynamically monitor the response times and load which are displayed in a series of graphs.
The users here are the number of people who control simultaneous connections to Oracle. We know that each user who connects to Oracle will be assigned the PGA, so this should be understood as the degree of parallelism.
2. Min/max Think times: the smallest/larger thinking time between each transaction. If you set min-do time, each transaction will be separated by a specified period.
3. Max Trans: If the setting will limit the maximum number of transactions.
4. The topmost transaction panel: Load:indicates the "weight" of the transaction in comparison to othe transactions. A higher weight indicates that it more likely to be run.
This panel is primarily able to cancel some of the fixed transaction types. Load This column is mainly used to adjust the weight of some transactions throughout the test. For example, browse product is primarily a SELECT statement that increases his weight and represents more people querying.
About several parameters in Oewizard:
Number of Customers: quantity of users preloaded into a database table.
Number of orders: orders that are preloaded into a database table.
The entire OE test is based on 9 sheets, so the test results are different by the amount of oewizard preloaded data.
For Oracle itself, indexed table performance is no different in size, but when the number of rows in a table reaches a certain level, such as a few billions of rows, the performance of the index is not as good as the full table scan. So for the range that OE allows, I don't think the table data size will affect performance very much.
Swingbench is a stress testing tool, and the result tpmc also represents the number of transactions per minute. If more data is preloaded and all types of weights in TX are fixed, the number of concurrent users needs to be adjusted to achieve an optimal TPMS value. The results of my previous test, the parallel user fixed, the number of preloaded data, the TPMC result is smaller, I was a bit confused, and then after careful analysis, I found that the number of concurrent users should be modified accordingly.
By modifying the weight of each transaction type in TX panel, you can also get some performance parameters of Oracle, such as query aggravation, if the value of TPMC is similar, the query ability of this database is good.

B,TPMC Introduction

According to TPC, the flow index describes how many New-order transactions can be processed per minute while the system performs the four transactions of payment, Order-status, Delivery and Stock-level. The response time for all transactions must meet the requirements of the TPC-C test specification.
The greater the flow indicator value, the better.
TPMC Calculation basis
In order to facilitate the calculation of database server modeling, we agreed:
"The system also has 1500 users online (U1);
"The average user sends 2 business requests per minute (N1);"
"The system sends out the business request, the update, the inquiry, the statistics each accounted for 1/3;"
"An average of 3 transactions per update business (T1);"
"An average of 8 transactions per enquiry Business (T2);"
"The average business generates 13 transactions per statistic (T3);"
"The amount of work in a day is 5 times times the average;
"Empirical coefficient is 1.6" (Practical Engineering Experience)
"Consider that the server retains 30% redundancy;
The processing capacity required by the server is:
Tpc-c=u1*n1* (T1+T2+T3)/3*3* empirical coefficient/redundancy coefficient
The processing performance of the application server is estimated as:
tpc-c= 1500*2* (3+8+13)/3*5*1.6/0.7= 274,285 tpmc
The database server is related to the stable operation of the whole system, considering high reliability and high availability, and focusing on the scalability and price/performance of the equipment, the system will configure two High-performance database servers with tpc-c values of no less than 280,000