mir-pucrs / smartix-rl

SmartIX

How to set up TPC-H Benchmark and train SmartIX

I. Requirements

• MySQL version 14.14 (see Section II)
• TPC-H version: 2.18.0 (see Section III)
• Python 3.6 (see Section V)

II. Installing MySQL

$ sudo apt-get install mysql-server
$ sudo mysql_secure_installation

VALIDATE PASSWORD PLUGIN (...) n
New password: YOUR PASSWORD
Re-enter new password: YOUR PASSWORD
Remove anonymous users? y
Disallow root login remotely? n
Remove test database and access to it? n
Reload privilege tables now? y

$ sudo ufw allow mysql
$ sudo systemctl start mysql
$ sudo systemctl enable mysql
$ sudo systemctl status mysql

Press Ctrl+C to exit MySQL status.

1. Create a new user:

$ sudo mysql -u root -p
mysql> CREATE USER 'smartix'@'localhost' IDENTIFIED BY 'smartix';
mysql> GRANT ALL PRIVILEGES ON *.* TO 'smartix'@'localhost' WITH GRANT OPTION;
mysql> CREATE USER 'smartix'@'%' IDENTIFIED BY 'smartix';
mysql> GRANT ALL PRIVILEGES ON *.* TO 'smartix'@'%' WITH GRANT OPTION;
mysql> FLUSH PRIVILEGES;
mysql> EXIT;

2. Create the database:

$ mysql -u smartix -p
mysql> CREATE DATABASE tpch;
mysql> EXIT;

III. Using TPC-H Tools

3. Click at TPC-H source code, available at the TPC website.

4. You have to fill in a register, and then they will send you a unique link by e-mail, so you can download and unzip the tools.

5. Go to the folder named dbgen, where there are the .c files.

6. Edit the makefile.suite file to match what we are showing below. We will use MySQL, but we have to set it to Oracle as the database here.

################
## CHANGE NAME OF ANSI COMPILER HERE
################
CC  	 = gcc
# Current values for DATABASE are: INFORMIX, DB2, TDAT (Teradata)
#                            SQLSERVER, SYBASE, ORACLE, VECTORWISE
# Current values for MACHINE are:  ATT, DOS, HP, IBM, ICL, MVS,
#                              	SGI, SUN, U2200, VMS, LINUX, WIN32
# Current values for WORKLOAD are:  TPCH
DATABASE = ORACLE
MACHINE  = LINUX
WORKLOAD = TPCH

7. Compile using the following command:

$ make -f makefile.suite

8. It will generate several compiled files. Now, we have to generate the database files. We use a scale factor of 1GB:

$ ./dbgen -s 1

9. It will generate one file for each table, to be inserted in the database, in the format tbl_name.tbl“.

10. Create the tables using the file dss.ddl:

$ mysql -u smartix -psmartix tpch < dss.ddl

11. Log in on MySQL with --local-infile variable enabled and select the TPCH database. Then, import the data to the database (pay attention to replace _~/path/ _for the actual path where the folder is):

mysql --local-infile=1 -u smartix -psmartix

USE tpch;
SET GLOBAL local_infile=1;
load data local infile '~/path/dbgen/region.tbl' into table REGION fields terminated by '|' lines terminated by '\n';
load data local infile '~/path/dbgen/nation.tbl' into table NATION fields terminated by '|' lines terminated by '\n';
load data local infile '~/path/dbgen/customer.tbl' into table CUSTOMER fields terminated by '|' lines terminated by '\n';
load data local infile '~/path/dbgen/supplier.tbl' into table SUPPLIER fields terminated by '|' lines terminated by '\n';
load data local infile '~/path/dbgen/part.tbl' into table PART fields terminated by '|' lines terminated by '\n';
load data local infile '~/path/dbgen/orders.tbl' into table ORDERS fields terminated by '|' lines terminated by '\n';
load data local infile '~/path/dbgen/partsupp.tbl' into table PARTSUPP fields terminated by '|' lines terminated by '\n';
load data local infile '~/path/dbgen/lineitem.tbl' into table LINEITEM fields terminated by '|' lines terminated by '\n';
exit;

12. Check the amount of rows for each table:

• REGION: 5
• NATION: 25
• CUSTOMER: 150000
• SUPPLIER: 10000
• PART: 200000
• ORDERS: 1500000
• PARTSUPP: 800000
• LINEITEM: 6001215

13. Generate the files to be used in refresh functions. It will generate files named delete.[num], and inserts to orders.[num], and lineitem.[num] tables. Create a folder named 1 (one) and move the files intto this folder. We do this in order to separate the refresh files created with respect to a scale factor of 1.

$ ./dbgen -s 1 -U 10000
$ mkdir 1
$ mv delete.* 1
$ mv orders.* 1
$ mv lineitem.* 1

IV. Configuring the TPC-H Benchmark

We need to create procedures, tables, and views to run our experiments.

14. Download the folder from this link and save it to your home folder.

15. Create folder /QSRF at /home/user/, and move files to this folder (/home/user/QSRF).

16. The files contained in the folder should be:

• README.txt: order to execute the files
• CREATE TEMPORARY TABLES.sql: create the tables to be used in data insertion and deletion;
• CREATE VIEWS.sh: create views to test database query perfomance;
• CREATE_PROCEDURE_DELETE_REFRESH_FUNCTION.sql: create procedure to delete data;
• CREATE_PROCEDURE_INSERT_REFRESH_FUNCTION.sql: create procedure to insert data;
• CREATE_PROCEDURE_QUERY_STREAM.sql: create procedure for query stream (sequential query in the 22 views we created).

17. Create temporary tables:

$ mysql -u dbuser -p tpch 
mysql> source ~/QSRF/CREATE TEMPORARY TABLES.sql;
mysql> exit;

18. Create views (check the path in the .sh file):

$ cd ~/QSRF
$ ./CREATE_VIEWS.sh

19. Create procedures:

$ mysql -u dbuser -p tpch < ~/QSRF/CREATE_PROCEDURE_DELETE_REFRESH_FUNCTION.sql;
$ mysql -u dbuser -p tpch < ~/QSRF/CREATE_PROCEDURE_INSERT_REFRESH_FUNCTION.sql;
$ mysql -u dbuser -p tpch < ~/QSRF/CREATE_PROCEDURE_QUERY_STREAM.sql;

20. Check if views and procedures are created:

$ mysql -u dbuser -p tpch 
mysql> use tpch;
mysql> show tables;

34 rows in set (0.00 sec)

mysql> show procedure status where db = 'tpch';

(...)

| tpch | DELETE_REFRESH_FUNCTION | PROCEDURE | dbuser@localhost | 
| tpch | INSERT_REFRESH_FUNCTION | PROCEDURE | dbuser@localhost | 
| tpch | QUERY_STREAM        	 | PROCEDURE | dbuser@localhost | 

(...)

3 rows in set (0.00 sec)

mysql> exit;

V. Configuring the Python Environment

• unixodbc ($ sudo apt-get install unixodbc-dev)
• pyodbc ($ pip3 install pyodbc)
• mysql-connector ($ pip3 install mysql-connector mysql-connector-python)

21. Download the MySQL Connector ODBC from this link.

VI. Training SmartIX

22. Download SmartIX source code (or clone the repository).

23. Configure the database connection string in the database.py class init method: put your user, password, and database name to the connection string variable.

24. The same has to be done in the TPCH.py class: put your database connection info to the DB_CONFIG constant, as well as setting the REFRESH_FILES_PATH constant to the path you generated the refresh files back in Step 13.

25. Then you can finally start training the agent by running:

$ python3 environment.py > training.log

36. Finally, you can view training data in the data folder.