Doris基础:修订间差异

来自牛奶河Wiki
跳到导航 跳到搜索
无编辑摘要
无编辑摘要
第14行: 第14行:
  distributed by hash(`ky`) buckets 1
  distributed by hash(`ky`) buckets 1
  properties (
  properties (
     "replication_allocation" = "tag.location.default: 1")
     "replication_allocation" = "tag.location.default: 1"
  ;
  )


===== Unique =====
===== Unique =====
第32行: 第32行:
  properties (
  properties (
     "replication_allocation" = "tag.location.default: 1",
     "replication_allocation" = "tag.location.default: 1",
     "enable_unique_key_merge_on_write" = "true")
     "enable_unique_key_merge_on_write" = "true"
  ;
  )
 
Duplicate


===== Duplicate =====
可重复模型
可重复模型
* 不对导入数据做任何操作
* 不对导入数据做任何操作
第49行: 第48行:
  properties (
  properties (
     "replication_allocation" = "tag.location.default: 1",
     "replication_allocation" = "tag.location.default: 1",
     "enable_duplicate_without_keys_by_default" = "true")
     "enable_duplicate_without_keys_by_default" = "true"
  ;
)
 
===== 分区、分桶 =====
* list
10 个分区,6 个桶,3 个副本
create table test_p
(
    part tinyint not null,
    ky  int,
    name varchar(10),
    val  int
)
duplicate key(part, ky)
partition by list(part)
  (
    partition p_0 values in(0),
    partition p_1 values in(1),
    partition p_2 values in(2),
    partition p_3 values in(3),
    partition p_4 values in(4),
    partition p_5 values in(5),
    partition p_6 values in(6),
    partition p_7 values in(7),
    partition p_8 values in(8),
    partition p_9 values in(9)
    -- partition p_0 values in(2,4,6,8,0),
    -- partition p_1 values in(1,3,5,7,9)
)
distributed by hash(ky) buckets 6
properties (
    "replication_allocation" = "tag.location.default: 3"
)
 
* range


[[分类:Develop]]
[[分类:Develop]]
[[分类:DB]]
[[分类:DB]]
[[分类:Doris]]
[[分类:Doris]]

2023年11月29日 (三) 16:40的版本

数据模型

Aggregate

聚合模型

  • Value 列会按照设置的 AggregationType 进行聚合,如:sum, max, replace 等
  • AGGREGATE KEY() 指定 key,未被指定的,需要提供 AggregationType,如:`cost` BIGINT SUM DEFAULT "0"
  • 读时合并(merge on read),因此在一些聚合查询上性能不佳
create table test_a
(
   ky   int,
   name varchar(10),
   val  int sum default "0" 
)
aggregate key(ky, name)
distributed by hash(`ky`) buckets 1
properties (
   "replication_allocation" = "tag.location.default: 1"
)
Unique

唯一模型

  • 保持 key 列的唯一,新值替换旧值
  • 写时合并(merge on write)
  • 可以在 be.conf 中添加配置项 disable_storage_page_cache=false,可能会优化数据导入性能
create table test_u
(
   ky   int,
   name varchar(10),
   val  int
)
unique key(ky, name)
distributed by hash(ky) buckets 1
properties (
   "replication_allocation" = "tag.location.default: 1",
   "enable_unique_key_merge_on_write" = "true"
)
Duplicate

可重复模型

  • 不对导入数据做任何操作
  • 建表语句中指定的 DUPLICATE KEY,只是用来指明底层数据按照那些列进行排序。(更贴切的名称应该为 “Sorted Column”)
create table test
(
   ky   int,
   name varchar(10),
   val  int
)
distributed by hash(ky) buckets 1
properties (
   "replication_allocation" = "tag.location.default: 1",
   "enable_duplicate_without_keys_by_default" = "true"
)
分区、分桶
  • list

10 个分区,6 个桶,3 个副本

create table test_p
(
   part tinyint not null,
   ky   int,
   name varchar(10),
   val  int
)
duplicate key(part, ky)
partition by list(part)
(
   partition p_0 values in(0),
   partition p_1 values in(1),
   partition p_2 values in(2),
   partition p_3 values in(3),
   partition p_4 values in(4),
   partition p_5 values in(5),
   partition p_6 values in(6),
   partition p_7 values in(7),
   partition p_8 values in(8),
   partition p_9 values in(9)
   -- partition p_0 values in(2,4,6,8,0),
   -- partition p_1 values in(1,3,5,7,9)
)
distributed by hash(ky) buckets 6
properties (
   "replication_allocation" = "tag.location.default: 3"
)
  • range