ClickHouse/tests/queries/0_stateless/00804_test_delta_codec_compression.sql

118 lines
4.1 KiB
MySQL
Raw Normal View History

2021-09-12 12:35:27 +00:00
-- Tags: no-parallel
2020-06-20 11:29:01 +00:00
SET send_logs_level = 'fatal';
SET joined_subquery_requires_alias = 0;
2019-06-07 16:02:24 +00:00
DROP TABLE IF EXISTS delta_codec_synthetic;
DROP TABLE IF EXISTS default_codec_synthetic;
2019-06-07 16:02:24 +00:00
CREATE TABLE delta_codec_synthetic
(
2020-03-24 16:53:12 +00:00
id UInt64 Codec(Delta, ZSTD(3))
2020-06-27 21:18:27 +00:00
) ENGINE MergeTree() ORDER BY tuple() SETTINGS min_bytes_for_wide_part = 0;
2019-06-07 16:02:24 +00:00
CREATE TABLE default_codec_synthetic
(
2020-03-24 16:53:12 +00:00
id UInt64 Codec(ZSTD(3))
2020-06-27 21:18:27 +00:00
) ENGINE MergeTree() ORDER BY tuple() SETTINGS min_bytes_for_wide_part = 0;
2019-06-07 16:02:24 +00:00
INSERT INTO delta_codec_synthetic SELECT number FROM system.numbers LIMIT 5000000;
INSERT INTO default_codec_synthetic SELECT number FROM system.numbers LIMIT 5000000;
2019-06-07 16:02:24 +00:00
OPTIMIZE TABLE delta_codec_synthetic FINAL;
OPTIMIZE TABLE default_codec_synthetic FINAL;
SELECT
floor(big_size / small_size) AS ratio
FROM
2020-11-13 13:25:18 +00:00
(SELECT 1 AS key, sum(bytes_on_disk) AS small_size FROM system.parts WHERE database == currentDatabase() and table == 'delta_codec_synthetic' and active)
2019-01-15 16:34:43 +00:00
INNER JOIN
2020-11-13 13:25:18 +00:00
(SELECT 1 AS key, sum(bytes_on_disk) as big_size FROM system.parts WHERE database == currentDatabase() and table == 'default_codec_synthetic' and active)
2019-01-15 16:34:43 +00:00
USING(key);
SELECT
small_hash == big_hash
FROM
2019-06-07 16:02:24 +00:00
(SELECT 1 AS key, sum(cityHash64(*)) AS small_hash FROM delta_codec_synthetic)
2019-01-15 16:34:43 +00:00
INNER JOIN
2019-06-07 16:02:24 +00:00
(SELECT 1 AS key, sum(cityHash64(*)) AS big_hash FROM default_codec_synthetic)
2019-01-15 16:34:43 +00:00
USING(key);
2019-06-07 16:02:24 +00:00
DROP TABLE IF EXISTS delta_codec_synthetic;
DROP TABLE IF EXISTS default_codec_synthetic;
2019-06-07 16:02:24 +00:00
DROP TABLE IF EXISTS delta_codec_float;
DROP TABLE IF EXISTS default_codec_float;
2019-06-07 16:02:24 +00:00
CREATE TABLE delta_codec_float
(
id Float64 Codec(Delta, LZ4HC)
2020-06-27 21:18:27 +00:00
) ENGINE MergeTree() ORDER BY tuple() SETTINGS min_bytes_for_wide_part = 0;
2019-06-07 16:02:24 +00:00
CREATE TABLE default_codec_float
(
id Float64 Codec(LZ4HC)
2020-06-27 21:18:27 +00:00
) ENGINE MergeTree() ORDER BY tuple() SETTINGS min_bytes_for_wide_part = 0;
2019-06-07 16:02:24 +00:00
INSERT INTO delta_codec_float SELECT number FROM numbers(1547510400, 500000) WHERE number % 3 == 0 OR number % 5 == 0 OR number % 7 == 0 OR number % 11 == 0;
INSERT INTO default_codec_float SELECT * from delta_codec_float;
2019-06-07 16:02:24 +00:00
OPTIMIZE TABLE delta_codec_float FINAL;
OPTIMIZE TABLE default_codec_float FINAL;
SELECT
floor(big_size / small_size) as ratio
FROM
2020-11-13 13:25:18 +00:00
(SELECT 1 AS key, sum(bytes_on_disk) AS small_size FROM system.parts WHERE database = currentDatabase() and table = 'delta_codec_float' and active)
INNER JOIN
2020-11-13 13:25:18 +00:00
(SELECT 1 AS key, sum(bytes_on_disk) as big_size FROM system.parts WHERE database = currentDatabase() and table = 'default_codec_float' and active) USING(key);
2019-01-15 16:34:43 +00:00
SELECT
small_hash == big_hash
FROM
2019-06-07 16:02:24 +00:00
(SELECT 1 AS key, sum(cityHash64(*)) AS small_hash FROM delta_codec_float)
2019-01-15 16:34:43 +00:00
INNER JOIN
2019-06-07 16:02:24 +00:00
(SELECT 1 AS key, sum(cityHash64(*)) AS big_hash FROM default_codec_float)
2019-01-15 16:34:43 +00:00
USING(key);
2019-06-07 16:02:24 +00:00
DROP TABLE IF EXISTS delta_codec_float;
DROP TABLE IF EXISTS default_codec_float;
2019-06-07 16:02:24 +00:00
DROP TABLE IF EXISTS delta_codec_string;
DROP TABLE IF EXISTS default_codec_string;
2019-06-07 16:02:24 +00:00
CREATE TABLE delta_codec_string
(
id Float64 Codec(Delta, LZ4)
2020-06-27 21:18:27 +00:00
) ENGINE MergeTree() ORDER BY tuple() SETTINGS min_bytes_for_wide_part = 0;
2019-06-07 16:02:24 +00:00
CREATE TABLE default_codec_string
(
id Float64 Codec(LZ4)
2020-06-27 21:18:27 +00:00
) ENGINE MergeTree() ORDER BY tuple() SETTINGS min_bytes_for_wide_part = 0;
2019-06-07 16:02:24 +00:00
INSERT INTO delta_codec_string SELECT concat(toString(number), toString(number % 100)) FROM numbers(1547510400, 500000);
INSERT INTO default_codec_string SELECT * from delta_codec_string;
2019-06-07 16:02:24 +00:00
OPTIMIZE TABLE delta_codec_string FINAL;
OPTIMIZE TABLE default_codec_string FINAL;
SELECT
floor(big_size / small_size) as ratio
FROM
2020-11-13 13:25:18 +00:00
(SELECT 1 AS key, sum(bytes_on_disk) AS small_size FROM system.parts WHERE database = currentDatabase() and table = 'delta_codec_string' and active)
INNER JOIN
2020-11-13 13:25:18 +00:00
(SELECT 1 AS key, sum(bytes_on_disk) as big_size FROM system.parts WHERE database = currentDatabase() and table = 'default_codec_string' and active) USING(key);
2019-01-15 16:34:43 +00:00
SELECT
small_hash == big_hash
FROM
2019-06-07 16:02:24 +00:00
(SELECT 1 AS key, sum(cityHash64(*)) AS small_hash FROM delta_codec_string)
2019-01-15 16:34:43 +00:00
INNER JOIN
2019-06-07 16:02:24 +00:00
(SELECT 1 AS key, sum(cityHash64(*)) AS big_hash FROM default_codec_string)
2019-01-15 16:34:43 +00:00
USING(key);
2019-06-07 16:02:24 +00:00
DROP TABLE IF EXISTS delta_codec_string;
DROP TABLE IF EXISTS default_codec_string;