Optimized and MapJoin friendly Q16 (#3962)

Author: Hor911

ydb/library/benchmarks/queries/tpch/ydb/q16.sql

@@ -2,42 +2,61 @@
 -- TPC TPC-H Parameter Substitution (Version 2.17.2 build 0)
 -- using 1680793381 as a seed to the RNG
 
-$join = (
+$p = (
 select
-    ps.ps_suppkey as ps_suppkey,
-    ps.ps_partkey as ps_partkey
+    p.p_brand as p_brand,
+    p.p_type as p_type,
+    p.p_size as p_size,
+    ps.ps_suppkey as ps_suppkey
 from
+    `{path}part` as p
+join
     `{path}partsupp` as ps
-left join
-    `{path}supplier` as w
 on
-    w.s_suppkey = ps.ps_suppkey
-where not (s_comment like "%Customer%Complaints%")
+    p.p_partkey = ps.ps_partkey
+where
+    p.p_brand <> 'Brand#33'
+    and (not StartsWith(p.p_type, 'PROMO POLISHED'))
+    and (p.p_size = 20 or p.p_size = 27 or p.p_size = 11 or p.p_size = 45 or p.p_size = 40 or p.p_size = 41 or p.p_size = 34 or p.p_size = 36)
+);
+
+$s = (
+select
+    s_suppkey
+from
+    `{path}supplier`
+where
+    s_comment like "%Customer%Complaints%"
 );
 
+$j = (
 select
     p.p_brand as p_brand,
     p.p_type as p_type,
     p.p_size as p_size,
-    count(distinct j.ps_suppkey) as supplier_cnt
+    p.ps_suppkey as ps_suppkey
 from
-    $join as j
-join
-    `{path}part` as p
+    $p as p
+left only join
+    $s as s
 on
-    p.p_partkey = j.ps_partkey
-where
-    p.p_brand <> 'Brand#33'
-    and (not StartsWith(p.p_type, 'PROMO POLISHED'))
-    and (p.p_size = 20 or p.p_size = 27 or p.p_size = 11 or p.p_size = 45 or p.p_size = 40 or p.p_size = 41 or p.p_size = 34 or p.p_size = 36)
+    p.ps_suppkey = s.s_suppkey
+);
+
+select
+    j.p_brand as p_brand,
+    j.p_type as p_type,
+    j.p_size as p_size,
+    count(distinct j.ps_suppkey) as supplier_cnt
+from
+    $j as j
 group by
-    p.p_brand,
-    p.p_type,
-    p.p_size
+    j.p_brand,
+    j.p_type,
+    j.p_size
 order by
     supplier_cnt desc,
     p_brand,
     p_type,
     p_size
 ;
-

ydb/library/benchmarks/queries/tpch/yql/q16.sql

@@ -4,42 +4,61 @@
 -- TPC TPC-H Parameter Substitution (Version 2.17.2 build 0)
 -- using 1680793381 as a seed to the RNG
 
-$join = (
+$p = (
 select
-    ps.ps_suppkey as ps_suppkey,
-    ps.ps_partkey as ps_partkey
+    p.p_brand as p_brand,
+    p.p_type as p_type,
+    p.p_size as p_size,
+    ps.ps_suppkey as ps_suppkey
 from
+    {{part}} as p
+join
     {{partsupp}} as ps
-left join
-    {{supplier}} as w
 on
-    w.s_suppkey = ps.ps_suppkey
-where not (s_comment like "%Customer%Complaints%")
+    p.p_partkey = ps.ps_partkey
+where
+    p.p_brand <> 'Brand#33'
+    and (not StartsWith(p.p_type, 'PROMO POLISHED'))
+    and (p.p_size = 20 or p.p_size = 27 or p.p_size = 11 or p.p_size = 45 or p.p_size = 40 or p.p_size = 41 or p.p_size = 34 or p.p_size = 36)
+);
+
+$s = (
+select
+    s_suppkey
+from
+    {{supplier}}
+where
+    s_comment like "%Customer%Complaints%"
 );
 
+$j = (
 select
     p.p_brand as p_brand,
     p.p_type as p_type,
     p.p_size as p_size,
-    count(distinct j.ps_suppkey) as supplier_cnt
+    p.ps_suppkey as ps_suppkey
 from
-    $join as j
-join
-    {{part}} as p
+    $p as p
+left only join
+    $s as s
 on
-    p.p_partkey = j.ps_partkey
-where
-    p.p_brand <> 'Brand#33'
-    and (not StartsWith(p.p_type, 'PROMO POLISHED'))
-    and (p.p_size = 20 or p.p_size = 27 or p.p_size = 11 or p.p_size = 45 or p.p_size = 40 or p.p_size = 41 or p.p_size = 34 or p.p_size = 36)
+    p.ps_suppkey = s.s_suppkey
+);
+
+select
+    j.p_brand as p_brand,
+    j.p_type as p_type,
+    j.p_size as p_size,
+    count(distinct j.ps_suppkey) as supplier_cnt
+from
+    $j as j
 group by
-    p.p_brand,
-    p.p_type,
-    p.p_size
+    j.p_brand,
+    j.p_type,
+    j.p_size
 order by
     supplier_cnt desc,
     p_brand,
     p_type,
     p_size
 ;
-