test: Add CH-benCHmark queries (#5600)

* add q02 * add q02 and q19 * remove tpcc_ prefix * add q12 * Add q11 and q17 * commit * add transactions * add q03 * add q04 * remove * add more queries * fix * create output for q17 Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
risingwavelabs · Oct 18, 2022 · f4e4475 · f4e4475
1 parent 8a650e5
commit f4e4475
Show file tree

Hide file tree

Showing 34 changed files with 6,391 additions and 1,015 deletions.
diff --git a/e2e_test/ch-benchmark/ch_benchmark.slt b/e2e_test/ch-benchmark/ch_benchmark.slt
@@ -1,13 +1,66 @@
-include ./create_sources.slt.part
+# Create tpch tables
 include ../tpch/create_tables.slt.part
 
+# Drop unneeded tables
+statement ok
+DROP TABLE lineitem;
+
+statement ok
+DROP TABLE orders;
+
+statement ok
+DROP TABLE customer;
+
+statement ok
+DROP TABLE partsupp;
+
+statement ok
+DROP TABLE part;
+
+# Insert data to tpch tables
+include ../tpch/insert_nation.slt.part
+include ../tpch/insert_region.slt.part
+include ../tpch/insert_supplier.slt.part
+
+# Create materialized sources from tpcc tables
+include ./create_sources.slt.part
+
 # Ensure that the upstream data is fully consumed
 statement ok
 flush;
 
-# include ./q01.slt.part --> precision error
+# Run CH-benCHmark queries
+include ./q01.slt.part
+include ./q02.slt.part
+include ./q03.slt.part
+include ./q04.slt.part
+# include ./q05.slt.part --> no output
+include ./q06.slt.part
+# include ./q07.slt.part --> no output
+# include ./q08.slt.part --> no output
+include ./q09.slt.part
+include ./q10.slt.part
+include ./q11.slt.part
+include ./q12.slt.part
 include ./q13.slt.part
+# include ./q14.slt.part --> panic, see issue: https://github.com/risingwavelabs/risingwave/issues/5563
+include ./q15.slt.part
 include ./q16.slt.part
+include ./q17.slt.part
+include ./q18.slt.part
+# include ./q19.slt.part --> precision error
+include ./q20.slt.part
+include ./q21.slt.part
+include ./q22.slt.part
 
 include ./drop_sources.slt.part
-include ../tpch/drop_tables.slt.part
+
+# Drop tpch tables
+statement ok
+DROP TABLE supplier;
+
+statement ok
+DROP TABLE region;
+
+statement ok
+DROP TABLE  nation;
diff --git a/e2e_test/ch-benchmark/create_sources.slt.part b/e2e_test/ch-benchmark/create_sources.slt.part
@@ -1,5 +1,5 @@
 statement ok
-CREATE MATERIALIZED SOURCE tpcc_customer (
+CREATE MATERIALIZED SOURCE customer (
     c_id INTEGER,
     c_d_id INTEGER,
     c_w_id INTEGER,
@@ -25,13 +25,13 @@ CREATE MATERIALIZED SOURCE tpcc_customer (
     PRIMARY KEY (c_id, c_d_id, c_w_id)
 ) with (
     connector = 'kafka',
-    topic = 'tpcc_customer',
+    topic = 'customer',
     properties.bootstrap.server = '127.0.0.1:29092',
     scan.startup.mode = 'earliest'
 ) ROW FORMAT DEBEZIUM_JSON;
 
 statement ok
-CREATE MATERIALIZED SOURCE tpcc_district (
+CREATE MATERIALIZED SOURCE district (
     d_id INTEGER,
     d_w_id INTEGER,
     d_name VARCHAR,
@@ -46,13 +46,13 @@ CREATE MATERIALIZED SOURCE tpcc_district (
     PRIMARY KEY (d_id, d_w_id)
 ) with (
     connector = 'kafka',
-    topic = 'tpcc_district',
+    topic = 'district',
     properties.bootstrap.server = '127.0.0.1:29092',
     scan.startup.mode = 'earliest'
 ) ROW FORMAT DEBEZIUM_JSON;
 
 # statement ok
-# CREATE MATERIALIZED SOURCE tpcc_history (
+# CREATE MATERIALIZED SOURCE history (
 #     h_c_id INTEGER,
 #     h_c_d_id INTEGER,
 #     h_c_w_id INTEGER,
@@ -63,13 +63,13 @@ CREATE MATERIALIZED SOURCE tpcc_district (
 #     h_data VARCHAR
 # ) with (
 #     connector = 'kafka',
-#     topic = 'tpcc_history',
+#     topic = 'history',
 #     properties.bootstrap.server = '127.0.0.1:29092',
 #     scan.startup.mode = 'earliest'
 # ) ROW FORMAT DEBEZIUM_JSON;
 
 statement ok
-CREATE MATERIALIZED SOURCE tpcc_item (
+CREATE MATERIALIZED SOURCE item (
     i_id INTEGER,
     i_im_id INTEGER,
     i_name VARCHAR,
@@ -78,26 +78,26 @@ CREATE MATERIALIZED SOURCE tpcc_item (
     PRIMARY KEY (i_id)
 ) with (
     connector = 'kafka',
-    topic = 'tpcc_item',
+    topic = 'item',
     properties.bootstrap.server = '127.0.0.1:29092',
     scan.startup.mode = 'earliest'
 ) ROW FORMAT DEBEZIUM_JSON;
 
 statement ok
-CREATE MATERIALIZED SOURCE tpcc_neworder (
+CREATE MATERIALIZED SOURCE neworder (
     no_o_id INTEGER,
     no_d_id INTEGER,
     no_w_id INTEGER,
     PRIMARY KEY (no_o_id, no_d_id, no_w_id)
 ) with (
     connector = 'kafka',
-    topic = 'tpcc_neworder',
+    topic = 'neworder',
     properties.bootstrap.server = '127.0.0.1:29092',
     scan.startup.mode = 'earliest'
 ) ROW FORMAT DEBEZIUM_JSON;
 
 statement ok
-CREATE MATERIALIZED SOURCE tpcc_orderline (
+CREATE MATERIALIZED SOURCE orderline (
     ol_o_id INTEGER,
     ol_d_id INTEGER,
     ol_w_id INTEGER,
@@ -111,13 +111,13 @@ CREATE MATERIALIZED SOURCE tpcc_orderline (
     PRIMARY KEY (ol_o_id, ol_d_id, ol_w_id, ol_number)
 ) with (
     connector = 'kafka',
-    topic = 'tpcc_orderline',
+    topic = 'orderline',
     properties.bootstrap.server = '127.0.0.1:29092',
     scan.startup.mode = 'earliest'
 ) ROW FORMAT DEBEZIUM_JSON;
 
 statement ok
-CREATE MATERIALIZED SOURCE tpcc_orders (
+CREATE MATERIALIZED SOURCE orders (
     o_id INTEGER,
     o_d_id INTEGER,
     o_w_id INTEGER,
@@ -129,13 +129,13 @@ CREATE MATERIALIZED SOURCE tpcc_orders (
     PRIMARY KEY (o_id, o_d_id, o_w_id)
 ) with (
     connector = 'kafka',
-    topic = 'tpcc_orders',
+    topic = 'orders',
     properties.bootstrap.server = '127.0.0.1:29092',
     scan.startup.mode = 'earliest'
 ) ROW FORMAT DEBEZIUM_JSON;
 
 statement ok
-CREATE MATERIALIZED SOURCE tpcc_stock (
+CREATE MATERIALIZED SOURCE stock (
     st_i_id INTEGER,
     st_w_id INTEGER,
     st_quantity INTEGER,
@@ -156,13 +156,13 @@ CREATE MATERIALIZED SOURCE tpcc_stock (
     PRIMARY KEY (st_i_id, st_w_id)
 ) with (
     connector = 'kafka',
-    topic = 'tpcc_stock',
+    topic = 'stock',
     properties.bootstrap.server = '127.0.0.1:29092',
     scan.startup.mode = 'earliest'
 ) ROW FORMAT DEBEZIUM_JSON;
 
 statement ok
-CREATE MATERIALIZED SOURCE tpcc_warehouse (
+CREATE MATERIALIZED SOURCE warehouse (
     w_id INTEGER,
     w_name VARCHAR,
     w_street_1 VARCHAR,
@@ -175,7 +175,7 @@ CREATE MATERIALIZED SOURCE tpcc_warehouse (
     PRIMARY KEY (w_id)
 ) with (
     connector = 'kafka',
-    topic = 'tpcc_warehouse',
+    topic = 'warehouse',
     properties.bootstrap.server = '127.0.0.1:29092',
     scan.startup.mode = 'earliest'
 ) ROW FORMAT DEBEZIUM_JSON;
diff --git a/e2e_test/ch-benchmark/drop_sources.slt.part b/e2e_test/ch-benchmark/drop_sources.slt.part
@@ -1,26 +1,26 @@
 statement ok
-DROP SOURCE tpcc_customer;
+DROP SOURCE customer;
 
 statement ok
-DROP SOURCE tpcc_district;
+DROP SOURCE district;
 
 # statement ok
-# DROP SOURCE tpcc_history;
+# DROP SOURCE history;
 
 statement ok
-DROP SOURCE tpcc_item;
+DROP SOURCE item;
 
 statement ok
-DROP SOURCE tpcc_neworder;
+DROP SOURCE neworder;
 
 statement ok
-DROP SOURCE tpcc_orderline;
+DROP SOURCE orderline;
 
 statement ok
-DROP SOURCE tpcc_orders;
+DROP SOURCE orders;
 
 statement ok
-DROP SOURCE tpcc_stock;
+DROP SOURCE stock;
 
 statement ok
-DROP SOURCE tpcc_warehouse;
+DROP SOURCE warehouse;
diff --git a/e2e_test/ch-benchmark/q01.slt.part b/e2e_test/ch-benchmark/q01.slt.part
@@ -0,0 +1,17 @@
+query IIIIII rowsort
+select   ol_number,
+         sum(ol_quantity) as sum_qty,
+         sum(ol_amount) as sum_amount,
+         avg(ol_quantity) as avg_qty,
+         avg(ol_amount) as avg_amount,
+         count(*) as count_order
+from     orderline
+where    ol_delivery_d > '2007-01-02 00:00:00.000000'
+group by ol_number order by ol_number;
+----
+         1     600         84                              5                0.7         120
+         2     550          0                              5                  0         110
+         3     560      14739 5.0909090909090909090909090909  133.9909090909091         110
+         4     550       44.1                              5 0.4009090770374645         110
+         5     550          0                              5                  0         110
+         6     200          0                              5                  0          40
diff --git a/e2e_test/ch-benchmark/q02.slt.part b/e2e_test/ch-benchmark/q02.slt.part
@@ -0,0 +1,22 @@
+query ITTITTTT rowsort
+select 	 s_suppkey, s_name, n_name, i_id, i_name, s_address, s_phone, s_comment
+from	 item, supplier, stock, nation, region,
+	     (select st_i_id as m_i_id,
+		         min(st_quantity) as m_st_quantity
+	     from	 stock, supplier, nation, region
+	     where	 (st_w_id*st_i_id) % 10000=s_suppkey
+	 	         and s_nationkey=n_nationkey
+	 	         and n_regionkey=r_regionkey
+	 	         and r_name like 'EUROP%'
+	     group by st_i_id) m
+where 	 i_id = st_i_id
+	     and (st_w_id * st_i_id) % 10000 = s_suppkey
+	     and s_nationkey = n_nationkey
+	     and n_regionkey = r_regionkey
+	     and i_data like '%B'
+	     and r_name like 'EUROP%'
+	     and i_id=m_i_id
+	     and st_quantity = m_st_quantity
+order by n_name, s_name, i_id;
+----
+11  Supplier#000000011  GERMANY    11  t4ccklaRBqjFXhCnH  JfwTs,LZrV, M,9C  28-613-996-1505  y ironic packages. slyly ironic accounts affix furiously; ironically unusual excuses across the flu
diff --git a/e2e_test/ch-benchmark/q03.slt.part b/e2e_test/ch-benchmark/q03.slt.part
@@ -0,0 +1,36 @@
+query IIIIT rowsort
+select	 ol_o_id, ol_w_id, ol_d_id,
+		 sum(ol_amount) as revenue, o_entry_d
+from 	 customer, neworder, orders, orderline
+where 	 c_state like 'A%'
+		 and c_id = o_c_id
+		 and c_w_id = o_w_id
+		 and c_d_id = o_d_id
+		 and no_w_id = o_w_id
+		 and no_d_id = o_d_id
+		 and no_o_id = o_id
+		 and ol_w_id = o_w_id
+		 and ol_d_id = o_d_id
+		 and ol_o_id = o_id
+		 and o_entry_d > '2007-01-02 00:00:00.000000'
+group by ol_o_id, ol_w_id, ol_d_id, o_entry_d
+order by revenue desc, o_entry_d;
+----
+    2111        1        4       21  2015-11-22 00:00:00
+    2107        1        4        0  2015-11-22 00:00:00
+    2105        1        7        0  2015-11-22 00:00:00
+    2112        1        2        0  2015-11-22 00:00:00
+    2106        1        9        0  2015-11-22 00:00:00
+    2108        1        6        0  2015-11-22 00:00:00
+    2108        1        2        0  2015-11-22 00:00:00
+    2109        1        4        0  2015-11-22 00:00:00
+    2112        1        8        0  2015-11-22 00:00:00
+    2114        1        4        0  2015-11-22 00:00:00
+    2110        1        7        0  2015-11-22 00:00:00
+    2107        1        8        0  2015-11-22 00:00:00
+    2113        1        2        0  2015-11-22 00:00:00
+    2106        1        4        0  2015-11-22 00:00:00
+    2111        1        9        0  2015-11-22 00:00:00
+    2112        1        4        0  2015-11-22 00:00:00
+    2107        1        2        0  2015-11-22 00:00:00
+    2113        1        6        0  2015-11-22 00:00:00
diff --git a/e2e_test/ch-benchmark/q04.slt.part b/e2e_test/ch-benchmark/q04.slt.part
@@ -0,0 +1,23 @@
+query II rowsort
+select	o_ol_cnt, count(*) as order_count
+from	orders
+where	o_entry_d >= '2007-01-02 00:00:00.000000'
+	    and o_entry_d < '2017-01-02 00:00:00.000000'
+	    and exists (select *
+		            from orderline
+		            where o_id = ol_o_id
+		            and o_w_id = ol_w_id
+		            and o_d_id = ol_d_id
+		            and ol_delivery_d >= o_entry_d)
+group	by o_ol_cnt
+order	by o_ol_cnt;
+----
+ 5           10
+ 6           14
+ 7           14
+ 8           16
+ 9           16
+10           10
+11           12
+12            6
+13            2
diff --git a/e2e_test/ch-benchmark/q06.slt.part b/e2e_test/ch-benchmark/q06.slt.part
@@ -0,0 +1,8 @@
+query R rowsort
+select sum(ol_amount) as revenue
+from orderline
+where ol_delivery_d >= '1999-01-01 00:00:00.000000'
+     and ol_delivery_d < '2020-01-01 00:00:00.000000'
+     and ol_quantity between 1 and 100000;
+----
+5446.26
diff --git a/e2e_test/ch-benchmark/q09.slt.part b/e2e_test/ch-benchmark/q09.slt.part
@@ -0,0 +1,16 @@
+query TTR rowsort
+select	 n_name, extract(year from o_entry_d::timestamp) as l_year, sum(ol_amount) as sum_profit
+from	 item, stock, supplier, orderline, orders, nation
+where	 ol_i_id = st_i_id
+    	 and ol_supply_w_id = st_w_id
+    	 and (st_w_id * st_i_id) % 10000 = s_suppkey
+    	 and ol_w_id = o_w_id
+    	 and ol_d_id = o_d_id
+    	 and ol_o_id = o_id
+    	 and ol_i_id = i_id
+    	 and s_nationkey = n_nationkey
+    	 and i_data like '%BB'
+group by n_name, extract(year from o_entry_d::timestamp)
+order by n_name, l_year desc;
+----
+ GERMANY    2015           0