青梅分布

Question

有一个 table 随机分布

CREATE TABLE schema.table (
    col1 int4 NULL,
    col2 int4 NULL,
    col3 int4 NULL
)
WITH (
    appendonly=true,
    compresstype=zstd,
    orientation=column
)
DISTRIBUTED RANDOMLY;

我们需要以最佳方式（以最小偏差）将行分布在一个字段上。为此，我们可以创建测试 tables

CREATE TABLE schema.test_table (
    col_1 int4 NULL,
    col_2 int4 NULL,
    col_3 int4 NULL
)
WITH (
    appendonly=true,
    compresstype=zstd,
    orientation=column
)
DISTRIBUTED BY (col_i); 
INSERT INTO schema.test_table SELECT * FROM schema.table;

然后检查它们是否有偏差，例如通过

select * from gp_toolkit.gp_skew_coefficient('schema.test_table'::regclass);

问题是我们想在不创建测试 table 的情况下检查 table 是否存在偏差。可以这样做吗？如果可以，怎么做？

Answer 1

如果您真的不想创建新的 table，这样的方法会起作用，但我不知道有什么方法可以在不实际（重新）分发数据的情况下对分布进行建模。

foo=# create table foo(a int, b int, c int);
NOTICE:  Table doesn't have 'DISTRIBUTED BY' clause -- Using column named 'a' as the Greenplum Database data distribution key for this table.
HINT:  The 'DISTRIBUTED BY' clause determines the distribution of data. Make sure column(s) chosen are the optimal data distribution key to minimize skew.
CREATE TABLE

foo=# insert into foo values (generate_series(1,100), generate_series(101,200), generate_series(2001, 2100));
INSERT 0 100

foo=# select * from gp_toolkit.gp_skew_coefficient('public.foo'::regclass);
 skcoid |         skccoeff
--------+--------------------------
  76788 | 18.460769214742921763000
(1 row)

foo=# select gp_segment_id, count(*) from foo group by 1 order by 1;
 gp_segment_id | count
---------------+-------
             0 |    17
             1 |    18
             2 |    23
             3 |    17
             4 |    15
             5 |    10
(6 rows)


foo=# ALTER TABLE foo SET
foo-# WITH (REORGANIZE=true)
foo-# DISTRIBUTED BY (a);
ALTER TABLE
foo=# select * from gp_toolkit.gp_skew_coefficient('public.foo'::regclass);
 skcoid |         skccoeff
--------+--------------------------
  76788 | 18.460769214742921763000
(1 row)

foo=# select gp_segment_id, count(*) from foo group by 1 order by 1;
 gp_segment_id | count
---------------+-------
             0 |    21
             1 |    18
             2 |    12
             3 |    15
             4 |    18
             5 |    16
(6 rows)

foo=#

foo=# ALTER TABLE foo SET
WITH (REORGANIZE=true)
DISTRIBUTED BY (b);
ALTER TABLE
foo=# select * from gp_toolkit.gp_skew_coefficient('public.foo'::regclass);
 skcoid |         skccoeff
--------+--------------------------
  76788 | 27.011108825814611346000
(1 row)

foo=# select gp_segment_id, count(*) from foo group by 1 order by 1;
 gp_segment_id | count
---------------+-------
             0 |    12
             1 |    14
             2 |    20
             3 |    24
             4 |    16
             5 |    14
(6 rows)

foo=#


foo=# ALTER TABLE foo SET
WITH (REORGANIZE=true)
DISTRIBUTED BY (c);
ALTER TABLE
foo=# select * from gp_toolkit.gp_skew_coefficient('public.foo'::regclass);
 skcoid |         skccoeff
--------+--------------------------
  76788 | 30.983866769659334938000
(1 row)

foo=#

foo=# select gp_segment_id, count(*) from foo group by 1 order by 1;
 gp_segment_id | count
---------------+-------
             0 |    19
             1 |    10
             2 |    20
             3 |    23
             4 |    11
             5 |    17
(6 rows)

foo=#

青梅分布

Greenplum distribution

sql

distribution

greenplum

skew