使用TPC-H测试oracle&infobright&postgresql的性能

• 下载tpch压缩包

http://www.tpc.org/tpch/

此网页的右中间有

• DBGEN & Reference Data Set zip (23 MB)

• 上传至linux并解压

[root@dn210120 software]# unzip tpch_2_14_3.zip

•  重命名makfile.suite为makefile并修改其中的内容

[root@dn210120 software]# cd dbgen/[root@dn210120 dbgen]# cp makefile.suite makefile[root@dn210120 dbgen]# vim makefile

修改成如下脚本：

################## 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:  TPCHDATABASE= MYSQLMACHINE = LINUXWORKLOAD = TPCH

• 增加mysql的脚本支持

[root@dn210120 dbgen]# vim tpcd.h

增加如下脚本：

#ifdef MYSQL#define GEN_QUERY_PLAN  ""#define START_TRAN      "START TRANSACTION"#define END_TRAN        "COMMIT"#define SET_OUTPUT      ""#define SET_ROWCOUNT    "limit %d;\n"#define SET_DBASE       "use %s;\n"#endif

• 编译tpch

[root@dn210120 dbgen]# make

会多了很多*.o文件，生成了dbgen和qgen可执行文件，生成了dss.ddl和dss.ri文件

• 生成数据脚本

[root@dn210120 dbgen]# ./dbgen -s 1

可通过查看README来了解其用法

• 修改dss.ddl和dss.ri文件

[root@dn210120 dbgen]# cat dss.ddl | tr A-Z a-z > dss-lower.ddl

[root@dn210120 dbgen]# cat dss.ri |tr A-Z a-z > dss-lower.ri[root@dn210120 dbgen]# sed -i "s/tpcd./tpch./g" dss-lower.ri[root@dn210120 dbgen]# sed -i "s/commit work/--commit work/g" dss-lower.ri[root@dn210120 dbgen]# sed -i "s/connect to tpch/--connect to tpch/g" dss-lower.ri

大小写转换、文本内容替换

• 创建数据库tpch

mysql> create database tpch;Query OK, 1 row affected (0.00 sec)

• 创建用户tpch

mysql> create user 'tpch'@'%' identified by 'tpch';Query OK, 0 rows affected (0.00 sec)

• 赋予用户tpch权限

mysql> grant all on tpch.* to 'tpch'@'%';Query OK, 0 rows affected (0.00 sec)

• 切换至数据库tpch并执行dss-lower.ddl 以便创建表

mysql> use tpch;Database changedmysql> \. ~/software/dbgen/dss-lower.ddlQuery OK, 0 rows affected (0.02 sec)Query OK, 0 rows affected (0.00 sec)Query OK, 0 rows affected (0.02 sec)Query OK, 0 rows affected (0.00 sec)Query OK, 0 rows affected (0.01 sec)Query OK, 0 rows affected (0.00 sec)Query OK, 0 rows affected (0.01 sec)Query OK, 0 rows affected (0.00 sec)mysql> show tables;+----------------+| Tables_in_tpch |+----------------+| customer       || lineitem       || nation         || orders         || part           || partsupp       || region         || supplier       |+----------------+8 rows in set (0.00 sec)

• 在数据库tpch下导入数据

mysql> LOAD DATA LOCAL INFILE '~/software/dbgen/customer.tbl' INTO TABLE customer FIELDS TERMINATED BY '|';Query OK, 150000 rows affected (1.76 sec)Records: 150000  Deleted: 0  Skipped: 0  Warnings: 0mysql> LOAD DATA LOCAL INFILE '~/software/dbgen/lineitem.tbl' INTO TABLE lineitem FIELDS TERMINATED BY '|';Query OK, 6001215 rows affected (1 min 12.30 sec)Records: 6001215  Deleted: 0  Skipped: 0  Warnings: 0mysql> LOAD DATA LOCAL INFILE '~/software/dbgen/nation.tbl' INTO TABLE nation FIELDS TERMINATED BY '|';Query OK, 25 rows affected (0.00 sec)Records: 25  Deleted: 0  Skipped: 0  Warnings: 0mysql> LOAD DATA LOCAL INFILE '~/software/dbgen/orders.tbl' INTO TABLE orders FIELDS TERMINATED BY '|';Query OK, 1500000 rows affected (14.87 sec)Records: 1500000  Deleted: 0  Skipped: 0  Warnings: 0mysql> LOAD DATA LOCAL INFILE '~/software/dbgen/part.tbl' INTO TABLE part FIELDS TERMINATED BY '|';Query OK, 200000 rows affected (2.31 sec)Records: 200000  Deleted: 0  Skipped: 0  Warnings: 0mysql> LOAD DATA LOCAL INFILE '~/software/dbgen/partsupp.tbl' INTO TABLE partsupp FIELDS TERMINATED BY '|';Query OK, 800000 rows affected (8.26 sec)Records: 800000  Deleted: 0  Skipped: 0  Warnings: 0mysql> LOAD DATA LOCAL INFILE '~/software/dbgen/region.tbl' INTO TABLE region FIELDS TERMINATED BY '|';Query OK, 5 rows affected (0.01 sec)Records: 5  Deleted: 0  Skipped: 0  Warnings: 0mysql> LOAD DATA LOCAL INFILE '~/software/dbgen/supplier.tbl' INTO TABLE supplier FIELDS TERMINATED BY '|';Query OK, 10000 rows affected (0.15 sec)Records: 10000  Deleted: 0  Skipped: 0  Warnings: 0

• 修改dss-lower.ri以致于匹配到mysql的语法标准

注意mysql的注释：--后要空格1.注释掉：commit work;2.注释掉：connect to tpch;

add foreign key nation_fk1 (n_regionkey) references tpch.region;=>add foreign key nation_fk1 (n_regionkey) references tpch.region(r_regionkey);

add foreign key supplier_fk1 (s_nationkey) references tpch.nation;=>add foreign key supplier_fk1 (s_nationkey) references tpch.nation(n_nationkey);

add foreign key customer_fk1 (c_nationkey) references tpch.nation;=>add foreign key customer_fk1 (c_nationkey) references tpch.nation(n_nationkey);

add foreign key partsupp_fk1 (ps_suppkey) references tpch.supplier; =>add foreign key partsupp_fk1 (ps_suppkey) references tpch.supplier(s_suppkey);

add foreign key partsupp_fk2 (ps_partkey) references tpch.part;=>add foreign key partsupp_fk2 (ps_partkey) references tpch.part(p_partkey);

add foreign key orders_fk1 (o_custkey) references tpch.customer;=>add foreign key orders_fk1 (o_custkey) references tpch.customer(c_custkey);

add foreign key lineitem_fk1 (l_orderkey)  references tpch.orders;=>add foreign key lineitem_fk1 (l_orderkey)  references tpch.orders(o_orderkey);

add foreign key lineitem_fk2 (l_partkey,l_suppkey) references        tpch.partsupp;=>add foreign key lineitem_fk2 (l_partkey,l_suppkey) references        tpch.partsupp(ps_partkey,ps_suppkey);

注意：一旦表中的数据依赖关系不全时，建立外键索引会报错

• 在数据库tpch下执行dss-lower.ri 以便建立表之间的关联关系

mysql> use tpch;Database changedmysql> \. ~/software/dbgen/dss-lower.ri

先插入数据再建索引，效率高

• 获取22条查询语句

[root@dn210120 dbgen]# cp ./queries*.sql ./[root@dn210120 dbgen]# ./qgen -s 1 > query.sql

• 修改query.sql中的sql

##去除 ^M 字符[root@dn210120 queries]# cat tpch_query.sql | col -b > tpch_query_new.sql

:%s/;\nlimit/ limit/g:%s/limit -1/limit 1/g:wq

##将 (3) 去除l_shipdate <= date '1998-12-01' - interval '77' day (3)

count(o_orderkey)=>count(o_orderkey) as c_count

) as c_orders (c_custkey, c_count)=>) as c_orders

drop view revenue0 limit 1;=>drop view revenue0;

• 将每条sql语句单独以文件保存

脚本执行测试有如下方法：

1. 自行开发测试工具
2. 研究tpchDriver测试工具
3. 结合shell脚本，手工执行脚本进行测试统计

测试使用到的软件：

1. jmeter
2. nmon/ganglia
3. tpch(dbgen、qgen)
4. oracle11.0.2
5. infobright 4.5
6. postgresql 9.3

测试维度：

1. 单用户
2. 多用户
3. 数据量级(1GB、10GB、30GB、100GB、300GB、1000GB)

测试方法：

1. 搭建数据库
2. 创建表
3. 导入数据
4. 建立索引
5. 单步顺序执行22条sql+2条dml
6. 获取单步执行响应时间
7. 并发执行随机选取的N条sql
8. 获取并发执行响应时间
9. 格式化nmon数据形成折线图或者直接使用ganglia进行查阅历史消耗资源情况

测试中用到的命令：

• oracle

sqlloader略

• infobright

mysql -h172.*.*.* -P5029 --local-infile=1 -utpch -ptpch tpch

LOAD DATA LOCAL INFILE '/data/wonhigh/data_scale/1/customer.tbl' INTO TABLE CUSTOMER FIELDS TERMINATED BY '|';
LOAD DATA LOCAL INFILE '/data/wonhigh/data_scale/1/orders.tbl'   INTO TABLE ORDERS   FIELDS TERMINATED BY '|';
LOAD DATA LOCAL INFILE '/data/wonhigh/data_scale/1/lineitem.tbl' INTO TABLE LINEITEM FIELDS TERMINATED BY '|';
LOAD DATA LOCAL INFILE '/data/wonhigh/data_scale/1/nation.tbl'   INTO TABLE NATION   FIELDS TERMINATED BY '|';
LOAD DATA LOCAL INFILE '/data/wonhigh/data_scale/1/partsupp.tbl' INTO TABLE PARTSUPP FIELDS TERMINATED BY '|';
LOAD DATA LOCAL INFILE '/data/wonhigh/data_scale/1/part.tbl'     INTO TABLE PART     FIELDS TERMINATED BY '|';
LOAD DATA LOCAL INFILE '/data/wonhigh/data_scale/1/region.tbl'   INTO TABLE REGION   FIELDS TERMINATED BY '|';
LOAD DATA LOCAL INFILE '/data/wonhigh/data_scale/1/supplier.tbl' INTO TABLE SUPPLIER FIELDS TERMINATED BY '|';

• postgresql

time psql -h172.17.209.102 -Utpch -c "COPY nation   FROM stdin WITH (FORMAT csv, DELIMITER '|')" < /data/wonhigh/data_scale/1/nation.tbl
time psql -h172.17.209.102 -Utpch -c "COPY region   FROM stdin WITH (FORMAT csv, DELIMITER '|')" < /data/wonhigh/data_scale/1/region.tbl
time psql -h172.17.209.102 -Utpch -c "COPY part     FROM stdin WITH (FORMAT csv, DELIMITER '|')" < /data/wonhigh/data_scale/1/part.tbl
time psql -h172.17.209.102 -Utpch -c "COPY supplier FROM stdin WITH (FORMAT csv, DELIMITER '|')" < /data/wonhigh/data_scale/1/supplier.tbl
time psql -h172.17.209.102 -Utpch -c "COPY partsupp FROM stdin WITH (FORMAT csv, DELIMITER '|')" < /data/wonhigh/data_scale/1/partsupp.tbl
time psql -h172.17.209.102 -Utpch -c "COPY customer FROM stdin WITH (FORMAT csv, DELIMITER '|')" < /data/wonhigh/data_scale/1/customer.tbl
time psql -h172.17.209.102 -Utpch -c "COPY orders   FROM stdin WITH (FORMAT csv, DELIMITER '|')" < /data/wonhigh/data_scale/1/orders.tbl
time psql -h172.17.209.102 -Utpch -c "COPY lineitem FROM stdin WITH (FORMAT csv, DELIMITER '|')" < /data/wonhigh/data_scale/1/lineitem.tbl

测试报告：

• 响应时间上

oracle的响应时间是三者中最快的，postgresql稍微略胜infobright

• 资源消耗上

oracle的在cpu、内存、IO上的使用没有太大的波动

infobright在cpu、IO上消耗波动比较大，内存的消耗上稍微小于postgresql

postgresql在大数据量时容易吃爆内存