iceberg native

Author: PingLiuPing

benchmarks/tpch-load/create-schema/sf100.sql

@@ -1,3 +1,3 @@
-CREATE SCHEMA IF NOT EXISTS load_tpch_sf1000_parquet WITH (LOCATION = 's3a://presto-workload/hive_java_load_tpch_sf1000_parquet/');
+CREATE SCHEMA IF NOT EXISTS load_tpch_sf1000_parquet WITH (LOCATION = 's3a://presto-workload/hive_native_load_tpch_sf1000_parquet/');
 --Do not need to explicitly speficy schema here
 USE load_tpch_sf1000_parquet;

benchmarks/tpch-load/hive-tpch-parquet-n-1000/customer_sf1000.sql

@@ -0,0 +1,12 @@
+-- Inserting into customer
+INSERT INTO customer
+SELECT
+    c_custkey,
+    CAST(c_name AS VARCHAR(25)),
+    CAST(c_address AS VARCHAR(40)),
+    c_nationkey,
+    CAST(c_phone AS VARCHAR(15)),
+    CAST(c_acctbal AS DECIMAL(12,2)),
+    CAST(c_mktsegment AS VARCHAR(10)),
+    CAST(c_comment AS VARCHAR(117))
+FROM tpch.sf1000.customer;

benchmarks/tpch-load/hive-tpch-parquet-n-1000/lineitem_sf1000.sql

@@ -0,0 +1,20 @@
+-- Inserting into lineitem
+INSERT INTO lineitem
+SELECT
+    l_orderkey,
+    l_partkey,
+    l_suppkey,
+    l_linenumber,
+    CAST(l_quantity AS DECIMAL(12,2)),
+    CAST(l_extendedprice AS DECIMAL(12,2)),
+    CAST(l_discount AS DECIMAL(12,2)),
+    CAST(l_tax AS DECIMAL(12,2)),
+    CAST(l_returnflag AS VARCHAR(1)),
+    CAST(l_linestatus AS VARCHAR(1)),
+    l_shipdate,
+    l_commitdate,
+    l_receiptdate,
+    CAST(l_shipinstruct AS VARCHAR(25)),
+    CAST(l_shipmode AS VARCHAR(10)),
+    CAST(l_comment AS VARCHAR(44))
+FROM tpch.sf1000.lineitem;

benchmarks/tpch-load/hive-tpch-parquet-n-1000/nation_sf1000.sql

@@ -0,0 +1,8 @@
+-- Inserting into nation
+INSERT INTO nation
+SELECT
+    n_nationkey,
+    CAST(n_name AS VARCHAR(25)),
+    n_regionkey,
+    CAST(n_comment AS VARCHAR(152))
+FROM tpch.sf1000.nation;

benchmarks/tpch-load/hive-tpch-parquet-n-1000/orders_sf1000.sql

@@ -0,0 +1,13 @@
+-- Inserting into orders
+INSERT INTO orders
+SELECT
+    o_orderkey,
+    o_custkey,
+    CAST(o_orderstatus AS VARCHAR(1)),
+    CAST(o_totalprice AS DECIMAL(12,2)),
+    o_orderdate,
+    CAST(o_orderpriority AS VARCHAR(15)),
+    CAST(o_clerk AS VARCHAR(15)),
+    o_shippriority,
+    CAST(o_comment AS VARCHAR(79))
+FROM tpch.sf1000.orders;

benchmarks/tpch-load/hive-tpch-parquet-n-1000/part_sf1000.sql

@@ -0,0 +1,13 @@
+-- Inserting into part
+INSERT INTO part
+SELECT
+    p_partkey,
+    CAST(p_name AS VARCHAR(55)),
+    CAST(p_mfgr AS VARCHAR(25)),
+    CAST(p_brand AS VARCHAR(10)),
+    CAST(p_type AS VARCHAR(25)),
+    p_size,
+    CAST(p_container AS VARCHAR(10)),
+    CAST(p_retailprice AS DECIMAL(12,2)),
+    CAST(p_comment AS VARCHAR(23))
+FROM tpch.sf1000.part;

benchmarks/tpch-load/hive-tpch-parquet-n-1000/partsupp_sf1000.sql

@@ -0,0 +1,9 @@
+-- Inserting into partsupp
+INSERT INTO partsupp
+SELECT
+    ps_partkey,
+    ps_suppkey,
+    ps_availqty,
+    CAST(ps_supplycost AS DECIMAL(12,2)),
+    CAST(ps_comment AS VARCHAR(199))
+FROM tpch.sf1000.partsupp;

benchmarks/tpch-load/hive-tpch-parquet-n-1000/region_sf1000.sql

@@ -0,0 +1,7 @@
+-- Inserting into region
+INSERT INTO region
+SELECT
+    r_regionkey,
+    CAST(r_name AS VARCHAR(25)),
+    CAST(r_comment AS VARCHAR(152))
+FROM tpch.sf1000.region;

benchmarks/tpch-load/hive-tpch-parquet-n-1000/supplier_sf1000.sql

@@ -0,0 +1,11 @@
+-- Inserting into supplier
+INSERT INTO supplier
+SELECT
+    s_suppkey,
+    CAST(s_name AS VARCHAR(25)),
+    CAST(s_address AS VARCHAR(40)),
+    s_nationkey,
+    CAST(s_phone AS VARCHAR(15)),
+    CAST(s_acctbal AS DECIMAL(12,2)),
+    CAST(s_comment AS VARCHAR(101))
+FROM tpch.sf1000.supplier;

benchmarks/tpch-load/insert_sf1000_n.json

@@ -0,0 +1,27 @@
+{
+  "description": "Loading tpch with scaling factor 10 to hive parquet tables.",
+  "query_files": [
+    "hive-tpch-parquet-n-1000/customer_sf1000.sql",
+    "hive-tpch-parquet-n-1000/lineitem_sf1000.sql",
+    "hive-tpch-parquet-n-1000/nation_sf1000.sql",
+    "hive-tpch-parquet-n-1000/orders_sf1000.sql",
+    "hive-tpch-parquet-n-1000/part_sf1000.sql",
+    "hive-tpch-parquet-n-1000/partsupp_sf1000.sql",
+    "hive-tpch-parquet-n-1000/region_sf1000.sql",
+    "hive-tpch-parquet-n-1000/supplier_sf1000.sql"
+  ],
+  "expected_row_counts": {
+    "load_tpch_sf100_parquet": [
+      1,
+      1,
+      1,
+      1,
+      1,
+      1,
+      1,
+      1,
+      1,
+      1
+    ]
+  }
+}