当我在外部表上完成流连接与将连续视图加入到外部表时,我得到不同的结果。我期望相同的两个查询似乎不同。我的本地管道实例和 fdw 中的表之间的延迟对我的连续流连接有影响吗?我正在尝试根据外部表格的 id 聚合 rx_bytes 和 tx_bytes。
我正在使用最新的 mysql_fdw
https://github.com/EnterpriseDB/mysql_fdw
创建外部表
CREATE EXTENSION mysql_fdw; CREATE SERVER local_mysql FOREIGN DATA WRAPPER mysql_fdw OPTIONS (host 'localhost', port '3306', secure_auth 'true'); CREATE USER MAPPING FOR pipeline SERVER local_mysql OPTIONS(username 'username', password 'password'); CREATE FOREIGN TABLE "foo_instance" ( "id" bigint, "foo_id" bigint, ) SERVERlocal_mysql OPTIONS(dbname 'schema', table_name 'foo_instance');
插入 10 次后,我希望这两个查询产生相同的结果:
一个)
CREATE CONTINUOUS VIEW total_bytes
AS SELECT date_trunc('minute', time_stamp::timestamp) AS minute,
id::integer,
sum(tx_bytes::bigint) AS tx_bytes,
sum(rx_bytes::bigint) AS rx_bytes
FROM byte_count_stream GROUP BY minute, id;
SELECT minute, sum(tx_bytes) AS tx_bytes, sum(rx_bytes) AS rx_bytes, foo_id
FROM total_bytes JOIN foo_instance
ON total_bytes.id=foo_instance.id GROUP BY minute, foo_instance.foo_id;
minute | tx_bytes | rx_bytes | foo_id
---------------------+----------+----------+---------
2016-02-22 09:04:00 | 450 | 513 | 7939
2016-02-22 09:04:00 | 2762 | 2210 | 7940
2016-02-22 09:04:00 | 143 | 332 | 7941
2016-02-22 09:04:00 | 371 | 1042 | 7942
2016-02-22 09:04:00 | 865 | 987 | 7943
(5 rows)
b)
CREATE CONTINUOUS VIEW joined_foo_total_bytes
AS SELECT date_trunc('minute', byte_count_stream.time_stamp::timestamp) AS minute,
sum(byte_count_stream.tx_bytes::bigint) AS tx_bytes,
sum(byte_count_stream.rx_bytes::bigint) AS rx_bytes,
foo_instance.foo_id
FROM byte_count_stream JOIN foo_instance ON byte_count_stream.id::integer = foo_instance.id
GROUP BY minute, foo_instance.foo_id;
pipeline=# select * from joined_user_total_bytes;
minute | tx_bytes | rx_bytes | foo_id
---------------------+----------+----------+---------
2016-02-22 09:04:00 | 371 | 1042 | 7942
2016-02-22 09:04:00 | 143 | 332 | 7941
2016-02-22 09:04:00 | 865 | 987 | 7943
2016-02-22 09:04:00 | 2762 | 2210 | 7940
(4 rows)
显然结果是不一样的。我可以从连续视图到外部表进行连接,但更喜欢使用流连接。