南大通用GBase 8a相同数据在不同数字字段类型下的磁盘占用情况验证

发表于 作者 laozizhu
GBase 8a相同数字不同类型的存储区别

本文用于验证，相同的数字，分别存在在tinyint, smallint ,int 和 bigint 类型时，GBase 8a数据库集群在存储层的磁盘占用是否有实质性差异。结论是占用磁盘一样无差别。

提示：默认GBase8a数据库集群是开启压缩的（5，5）。

测试环境

2个节点的虚拟机

[gbase@localhost test_int_n1]$ gcadmin
CLUSTER STATE:  ACTIVE
CLUSTER MODE:   NORMAL

=================================================================
|             GBASE COORDINATOR CLUSTER INFORMATION             |
=================================================================
|   NodeName   |     IpAddress     |gcware |gcluster |DataState |
-----------------------------------------------------------------
| coordinator1 |    10.0.2.107     | OPEN  |  OPEN   |    0     |
-----------------------------------------------------------------
| coordinator2 |    10.0.2.106     | OPEN  |  OPEN   |    0     |
-----------------------------------------------------------------
=============================================================
|              GBASE DATA CLUSTER INFORMATION               |
=============================================================
|NodeName |     IpAddress     |gnode |syncserver |DataState |
-------------------------------------------------------------
|  node1  |    10.0.2.107     | OPEN |   OPEN    |    0     |
-------------------------------------------------------------
|  node2  |    10.0.2.106     | OPEN |   OPEN    |    0     |
-------------------------------------------------------------

测试表

包含tinyint, smallint, int 和bigint四个字段。

gbase> desc test_int;
+-------+-------------+------+-----+---------+-------+
| Field | Type        | Null | Key | Default | Extra |
+-------+-------------+------+-----+---------+-------+
| id    | tinyint(4)  | YES  |     | NULL    |       |
| id2   | smallint(6) | YES  |     | NULL    |       |
| id3   | int(11)     | YES  |     | NULL    |       |
| id4   | bigint(20)  | YES  |     | NULL    |       |
+-------+-------------+------+-----+---------+-------+
4 rows in set (Elapsed: 00:00:00.00)

测试数据

通过如下shell产生，每行相同的数字输出4次

[gbase@localhost ~]$ cat gen_int.sh
for((i=1;i<=10000000;i++))
do
  declare -i number=$RANDOM*100/32767;
  echo ${number},${number},${number},${number}
done
[gbase@localhost ~]$ sh gen_int.sh > 1000w.dat

第一次加载结果

可以看到，4个列数据部分占用的磁盘空间完全一样。

gbase> load data infile 'file://10.0.2.107/home/gbase/1000w.dat' into table test_int fields terminated by ',' ;
Query OK, 10000000 rows affected (Elapsed: 00:00:24.84)
Task 1835156 finished, Loaded 10000000 records, Skipped 0 records


[gbase@localhost test_int_n1]$ ll
total 32512
-rw------- 1 gbase gbase 5002613 Sep  4 21:41 C00000.seg.1
-rw------- 1 gbase gbase 5002613 Sep  4 21:41 C00001.seg.1
-rw------- 1 gbase gbase 5002613 Sep  4 21:41 C00002.seg.1
-rw------- 1 gbase gbase 5002613 Sep  4 21:41 C00003.seg.1
[gbase@localhost test_int_n1]$

[gbase@localhost test_int_n1]$ ll ../../metadata/test_int_n1.GED
total 80
-rw------- 1 gbase gbase  129 Sep  4 21:41 C00000.ctl.A
-rw------- 1 gbase gbase  129 Sep  4 20:59 C00000.ctl.B
-rw------- 1 gbase gbase 4134 Sep  4 21:41 C00000.map
-rw------- 1 gbase gbase  130 Sep  4 21:41 C00001.ctl.A
-rw------- 1 gbase gbase  130 Sep  4 20:59 C00001.ctl.B
-rw------- 1 gbase gbase 4134 Sep  4 21:41 C00001.map
-rw------- 1 gbase gbase  130 Sep  4 21:41 C00002.ctl.A
-rw------- 1 gbase gbase  130 Sep  4 20:59 C00002.ctl.B
-rw------- 1 gbase gbase 4134 Sep  4 21:41 C00002.map
-rw------- 1 gbase gbase  130 Sep  4 21:41 C00003.ctl.A
-rw------- 1 gbase gbase  130 Sep  4 20:59 C00003.ctl.B
-rw------- 1 gbase gbase 4134 Sep  4 21:41 C00003.map
-rw------- 1 gbase gbase  249 Sep  4 21:41 table.des.A
-rw------- 1 gbase gbase  249 Sep  4 20:59 table.des.B
-rw------- 1 gbase gbase   52 Sep  4 21:41 table.state.A
-rw------- 1 gbase gbase   52 Sep  4 20:59 table.state.B
[gbase@localhost test_int_n1]$

第二次加载

结果一致，A/B版本切换，新数据文件也一致。

和第一次略有差距，是因为DC包 65536行。上一个版本的不足65536行的尾块要复制到下一个版本里。

gbase> load data infile 'file://10.0.2.107/home/gbase/1000w.dat' into table test_int fields terminated by ',' ;
Query OK, 10000000 rows affected (Elapsed: 00:00:25.39)
Task 1835157 finished, Loaded 10000000 records, Skipped 0 records
gbase> select count(*) from test_int;
+----------+
| count(*) |
+----------+
| 20000000 |
+----------+
1 row in set (Elapsed: 00:00:00.01)

gbase> select * from test_int limit 10;
+------+------+------+------+
| id   | id2  | id3  | id4  |
+------+------+------+------+
|   52 |   52 |   52 |   52 |
|   74 |   74 |   74 |   74 |
|   51 |   51 |   51 |   51 |
|   60 |   60 |   60 |   60 |
|   22 |   22 |   22 |   22 |
|   38 |   38 |   38 |   38 |
|   23 |   23 |   23 |   23 |
|   21 |   21 |   21 |   21 |
|    5 |    5 |    5 |    5 |
|   17 |   17 |   17 |   17 |
+------+------+------+------+
10 rows in set (Elapsed: 00:00:00.01)


[gbase@localhost test_int_n1]$ ll
total 39184
-rw------- 1 gbase gbase 5021878 Sep  4 21:44 C00000.seg
-rw------- 1 gbase gbase 5002613 Sep  4 21:41 C00000.seg.1
-rw------- 1 gbase gbase 5021878 Sep  4 21:44 C00001.seg
-rw------- 1 gbase gbase 5002613 Sep  4 21:41 C00001.seg.1
-rw------- 1 gbase gbase 5021878 Sep  4 21:44 C00002.seg
-rw------- 1 gbase gbase 5002613 Sep  4 21:41 C00002.seg.1
-rw------- 1 gbase gbase 5021878 Sep  4 21:44 C00003.seg
-rw------- 1 gbase gbase 5002613 Sep  4 21:41 C00003.seg.1
[gbase@localhost test_int_n1]$
[gbase@localhost test_int_n1]$ ll ../../metadata/test_int_n1.GED
total 80
-rw------- 1 gbase gbase  129 Sep  4 21:41 C00000.ctl.A
-rw------- 1 gbase gbase  129 Sep  4 21:44 C00000.ctl.B
-rw------- 1 gbase gbase 8162 Sep  4 21:44 C00000.map
-rw------- 1 gbase gbase  130 Sep  4 21:41 C00001.ctl.A
-rw------- 1 gbase gbase  130 Sep  4 21:44 C00001.ctl.B
-rw------- 1 gbase gbase 8162 Sep  4 21:44 C00001.map
-rw------- 1 gbase gbase  130 Sep  4 21:41 C00002.ctl.A
-rw------- 1 gbase gbase  130 Sep  4 21:44 C00002.ctl.B
-rw------- 1 gbase gbase 8162 Sep  4 21:44 C00002.map
-rw------- 1 gbase gbase  130 Sep  4 21:41 C00003.ctl.A
-rw------- 1 gbase gbase  130 Sep  4 21:44 C00003.ctl.B
-rw------- 1 gbase gbase 8162 Sep  4 21:44 C00003.map
-rw------- 1 gbase gbase    0 Sep  4 21:44 lock
-rw------- 1 gbase gbase  249 Sep  4 21:41 table.des.A
-rw------- 1 gbase gbase  249 Sep  4 21:44 table.des.B
-rw------- 1 gbase gbase   52 Sep  4 21:41 table.state.A
-rw------- 1 gbase gbase   52 Sep  4 21:44 table.state.B
[gbase@localhost test_int_n1]$

第三次加载

一致

gbase> load data infile 'file://10.0.2.107/home/gbase/1000w.dat' into table test_int fields terminated by ',' ;
Query OK, 10000000 rows affected (Elapsed: 00:00:24.95)
Task 1835158 finished, Loaded 10000000 records, Skipped 0 records

gbase> select count(*) from test_int;
+----------+
| count(*) |
+----------+
| 30000000 |
+----------+
1 row in set (Elapsed: 00:00:00.01)



[gbase@localhost test_int_n1]$ ll
total 71856
-rw------- 1 gbase gbase  5021878 Sep  4 21:44 C00000.seg
-rw------- 1 gbase gbase 10024464 Sep  4 21:46 C00000.seg.1
-rw------- 1 gbase gbase  5021878 Sep  4 21:44 C00001.seg
-rw------- 1 gbase gbase 10024464 Sep  4 21:46 C00001.seg.1
-rw------- 1 gbase gbase  5021878 Sep  4 21:44 C00002.seg
-rw------- 1 gbase gbase 10024464 Sep  4 21:46 C00002.seg.1
-rw------- 1 gbase gbase  5021878 Sep  4 21:44 C00003.seg
-rw------- 1 gbase gbase 10024464 Sep  4 21:46 C00003.seg.1
[gbase@localhost test_int_n1]$
[gbase@localhost test_int_n1]$ ll ../../metadata/test_int_n1.GED
total 96
-rw------- 1 gbase gbase   129 Sep  4 21:46 C00000.ctl.A
-rw------- 1 gbase gbase   129 Sep  4 21:44 C00000.ctl.B
-rw------- 1 gbase gbase 12190 Sep  4 21:46 C00000.map
-rw------- 1 gbase gbase   130 Sep  4 21:46 C00001.ctl.A
-rw------- 1 gbase gbase   130 Sep  4 21:44 C00001.ctl.B
-rw------- 1 gbase gbase 12190 Sep  4 21:46 C00001.map
-rw------- 1 gbase gbase   130 Sep  4 21:46 C00002.ctl.A
-rw------- 1 gbase gbase   130 Sep  4 21:44 C00002.ctl.B
-rw------- 1 gbase gbase 12190 Sep  4 21:46 C00002.map
-rw------- 1 gbase gbase   130 Sep  4 21:46 C00003.ctl.A
-rw------- 1 gbase gbase   130 Sep  4 21:44 C00003.ctl.B
-rw------- 1 gbase gbase 12190 Sep  4 21:46 C00003.map
-rw------- 1 gbase gbase   249 Sep  4 21:46 table.des.A
-rw------- 1 gbase gbase   249 Sep  4 21:44 table.des.B
-rw------- 1 gbase gbase   52 Sep   4 21:46 table.state.A
-rw------- 1 gbase gbase   52 Sep   4 21:44 table.state.B
[gbase@localhost test_int_n1]$

结论

GBase 8a数据库，在对数据压缩时，不考虑字段定义，所以相同数据占用的磁盘空间是一样的。

Post Views: 482

相关文章:

  1. 南大通用GBase 8a集群版本命名规范
  2. 南大通用GBase 8a集群V95版本gcwar服务snapshot文件功能说明
  3. 南大通用GBase 8a集群通过sftp和ftp加载单个小文件非常慢的问题，关闭DNS反向解析
  4. 南大通用GBase 8a 数据库集群常用优化手段建议方法
  5. 南大通用GBase 8a数据库快速手工卸载
  6. 南大通用GBase 8a 加载LOAD数据无法入库错误 skipped records，text parser error
  7. 南大通用GBase 8a从V8升级到V9的REDOLOG的刷新处理
  8. 数据库SQL面试题：SQL做JOIN时，额外的过滤条件在where和on里的不同
  9. 南大通用GBase 8a节点数据迁移到其它存储盘阵的磁盘空间扩容方案
  10. 南大通用GBase 8a扩容时从新策略distribution回退到老策略的方案