change partitioning in tpch (#17416)

Co-authored-by: Nikolay Perfilov <pnv1@yandex-team.ru>

16 files changed, 828 insertions, 14 deletions

diff --git a/ydb/apps/ydb/CHANGELOG.md b/ydb/apps/ydb/CHANGELOG.md
index 637e78e4e2b..8fe9b9c2cdd 100644
--- a/ydb/apps/ydb/CHANGELOG.md
+++ b/ydb/apps/ydb/CHANGELOG.md
@@ -1,3 +1,4 @@
+* Added `--scale` option to `ydb workload tpch init` and `ydb workload tpcds init` commands. Sets the percentage of the benchmark's data size and workload to use, relative to full scale.
 * Added "--no-discovery" option. It allows to skip discovery and use user provided endpoint to connect to YDB cluster.
 * Added `--retries` to `ydb workload <clickbenh|tpch|tpcds> run` command.
 * Added `--partition-size` param to `ydb workload <clickbench/tpcds/tpch> init`.
diff --git a/ydb/library/workload/benchmark_base/workload.cpp b/ydb/library/workload/benchmark_base/workload.cpp
index 7e5d09aaffe..cb413f4cca0 100644
--- a/ydb/library/workload/benchmark_base/workload.cpp
+++ b/ydb/library/workload/benchmark_base/workload.cpp
@@ -35,6 +35,22 @@ const TString TWorkloadGeneratorBase::CsvFormatString = [] () {
     return settings.SerializeAsString();
 } ();
 
+namespace {
+
+    TString KeysList(const NJson::TJsonValue& table, const TString& key) {
+        TVector<TStringBuf> keysV;
+        for (const auto& k: table[key].GetArray()) {
+            keysV.emplace_back(k.GetString());
+        }
+        return JoinSeq(", ", keysV);
+    }
+
+}
+
+ui32 TWorkloadGeneratorBase::GetDefaultPartitionsCount(const TString& /*tableName*/) const {
+    return 64;
+}
+
 void TWorkloadGeneratorBase::GenerateDDLForTable(IOutputStream& result, const NJson::TJsonValue& table, bool single) const {
     auto specialTypes = GetSpecialDataTypes();
     specialTypes["string_type"] = Params.GetStringType();
@@ -65,11 +81,7 @@ void TWorkloadGeneratorBase::GenerateDDLForTable(IOutputStream& result, const NJ
         }
     }
     result << JoinSeq(",\n", columns);
-    TVector<TStringBuf> keysV;
-    for (const auto& k: table["primary_key"].GetArray()) {
-        keysV.emplace_back(k.GetString());
-    }
-    const TString keys = JoinSeq(", ", keysV);
+    const auto keys = KeysList(table, "primary_key");
     if (Params.GetStoreType() == TWorkloadBaseParams::EStoreType::ExternalS3) {
         result << Endl;
     } else {
@@ -78,7 +90,7 @@ void TWorkloadGeneratorBase::GenerateDDLForTable(IOutputStream& result, const NJ
     result << ")" << Endl;
 
     if (Params.GetStoreType() == TWorkloadBaseParams::EStoreType::Column) {
-        result << "PARTITION BY HASH (" << keys << ")" << Endl;
+        result << "PARTITION BY HASH (" <<  (table.Has("partition_by") ? KeysList(table, "partition_by") : keys) << ")" << Endl;
     }
 
     result << "WITH (" << Endl;
@@ -89,12 +101,12 @@ void TWorkloadGeneratorBase::GenerateDDLForTable(IOutputStream& result, const NJ
         break;
     case TWorkloadBaseParams::EStoreType::Column:
         result << "    STORE = COLUMN," << Endl;
-        result << "    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = " << table["partitioning"].GetUIntegerSafe(64) << Endl;
+        result << "    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = " << table["partitioning"].GetUIntegerSafe(GetDefaultPartitionsCount(tableName)) << Endl;
         break;
     case TWorkloadBaseParams::EStoreType::Row:
         result << "    STORE = ROW," << Endl;
         result << "    AUTO_PARTITIONING_PARTITION_SIZE_MB = " << Params.GetPartitionSizeMb() << ", " << Endl;
-        result << "    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = " << table["partitioning"].GetUIntegerSafe(64) << Endl;
+        result << "    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = " << table["partitioning"].GetUIntegerSafe(GetDefaultPartitionsCount(tableName)) << Endl;
     }
     result << ");" << Endl;
 }
diff --git a/ydb/library/workload/benchmark_base/workload.h b/ydb/library/workload/benchmark_base/workload.h
index 52cb0c2774c..6ddf1cbd917 100644
--- a/ydb/library/workload/benchmark_base/workload.h
+++ b/ydb/library/workload/benchmark_base/workload.h
@@ -53,6 +53,7 @@ protected:
     virtual TString GetTablesYaml() const = 0;
     virtual TSpecialDataTypes GetSpecialDataTypes() const = 0;
     NJson::TJsonValue GetTablesJson() const;
+    virtual ui32 GetDefaultPartitionsCount(const TString& tableName) const;
 
     THolder<TGeneratorStateProcessor> StateProcessor;
 private:
diff --git a/ydb/library/workload/tpc_base/tpc_base.cpp b/ydb/library/workload/tpc_base/tpc_base.cpp
index 1a91da20851..0e02ee556ee 100644
--- a/ydb/library/workload/tpc_base/tpc_base.cpp
+++ b/ydb/library/workload/tpc_base/tpc_base.cpp
@@ -167,7 +167,7 @@ void TTpcBaseWorkloadParams::ConfigureOpts(NLastGetopt::TOpts& opts, const EComm
             .StoreResult(&ExternalQueriesDir);
         opts.AddLongOption( "syntax", "Query syntax [" + GetEnumAllNames<EQuerySyntax>() + "].")
             .StoreResult(&Syntax).DefaultValue(Syntax);
-        opts.AddLongOption("scale", "scale in percents")
+        opts.AddLongOption("scale", "Sets the percentage of the benchmark's data size and workload to use, relative to full scale.")
             .DefaultValue(Scale).StoreResult(&Scale);
         opts.AddLongOption("float-mode", "Float mode. Can be float, decimal or decimal_ydb. If set to 'float' - float will be used, 'decimal' means that decimal will be used with canonical size and 'decimal_ydb' means that all floats will be converted to decimal(22,9) because YDB supports only this type.")
             .StoreResult(&FloatMode).DefaultValue(FloatMode);
@@ -177,6 +177,8 @@ void TTpcBaseWorkloadParams::ConfigureOpts(NLastGetopt::TOpts& opts, const EComm
     case TWorkloadParams::ECommandType::Init:
         opts.AddLongOption("float-mode", "Float mode. Can be float, decimal or decimal_ydb. If set to 'float' - float will be used, 'decimal' means that decimal will be used with canonical size and 'decimal_ydb' means that all floats will be converted to decimal(22,9) because YDB supports only this type.")
             .StoreResult(&FloatMode).DefaultValue(FloatMode);
+        opts.AddLongOption("scale", "Sets the percentage of the benchmark's data size and workload to use, relative to full scale.")
+            .DefaultValue(Scale).StoreResult(&Scale);
         break;
     default:
         break;
diff --git a/ydb/library/workload/tpcds/data_generator.cpp b/ydb/library/workload/tpcds/data_generator.cpp
index b15589ca550..495ba99f8d5 100644
--- a/ydb/library/workload/tpcds/data_generator.cpp
+++ b/ydb/library/workload/tpcds/data_generator.cpp
@@ -19,7 +19,7 @@ TTpcdsWorkloadDataInitializerGenerator::TTpcdsWorkloadDataInitializerGenerator(c
 
 void TTpcdsWorkloadDataInitializerGenerator::ConfigureOpts(NLastGetopt::TOpts& opts) {
     TWorkloadDataInitializerBase::ConfigureOpts(opts);
-    opts.AddLongOption("scale", "scale in percents")
+    opts.AddLongOption("scale", "Sets the percentage of the benchmark's data size and workload to use, relative to full scale.")
         .DefaultValue(Scale).StoreResult(&Scale);
     opts.AddLongOption("tables", "Commaseparated list of tables for generate. Empty means all tables.\n"
             "Enabled tables: " + JoinSeq(", ", TBulkDataGenerator::TFactory::GetRegisteredKeys()))
diff --git a/ydb/library/workload/tpch/data_generator.cpp b/ydb/library/workload/tpch/data_generator.cpp
index 1fdbe46142b..24e2f74e8a9 100644
--- a/ydb/library/workload/tpch/data_generator.cpp
+++ b/ydb/library/workload/tpch/data_generator.cpp
@@ -11,7 +11,7 @@ TTpchWorkloadDataInitializerGenerator::TTpchWorkloadDataInitializerGenerator(con
 
 void TTpchWorkloadDataInitializerGenerator::ConfigureOpts(NLastGetopt::TOpts& opts) {
     TWorkloadDataInitializerBase::ConfigureOpts(opts);
-    opts.AddLongOption("scale", "scale in percents")
+    opts.AddLongOption("scale", "Sets the percentage of the benchmark's data size and workload to use, relative to full scale.")
         .DefaultValue(Scale).StoreResult(&Scale);
     opts.AddLongOption("tables", "Commaseparated list of tables for generate. Empty means all tables.\n"
             "Enabled tables: " + JoinSeq(", ", TBulkDataGenerator::TFactory::GetRegisteredKeys()))
diff --git a/ydb/library/workload/tpch/tpch.cpp b/ydb/library/workload/tpch/tpch.cpp
index 183f3a0e7b1..13de9152902 100644
--- a/ydb/library/workload/tpch/tpch.cpp
+++ b/ydb/library/workload/tpch/tpch.cpp
@@ -29,6 +29,10 @@ TWorkloadGeneratorBase::TSpecialDataTypes TTpchWorkloadGenerator::GetSpecialData
     }
 }
 
+ui32 TTpchWorkloadGenerator::GetDefaultPartitionsCount(const TString& /*tableName*/) const {
+    return Params.GetScale() <= 10 ? 64 : 256;
+}
+
 
 THolder<IWorkloadQueryGenerator> TTpchWorkloadParams::CreateGenerator() const {
     return MakeHolder<TTpchWorkloadGenerator>(*this);
diff --git a/ydb/library/workload/tpch/tpch.h b/ydb/library/workload/tpch/tpch.h
index bfe5506c493..7e801ba73ec 100644
--- a/ydb/library/workload/tpch/tpch.h
+++ b/ydb/library/workload/tpch/tpch.h
@@ -19,6 +19,7 @@ public:
 protected:
     TString GetTablesYaml() const override;
     TWorkloadGeneratorBase::TSpecialDataTypes GetSpecialDataTypes() const override;
+    ui32 GetDefaultPartitionsCount(const TString& tableName) const override;
 
 private:
     const TTpchWorkloadParams& Params;
diff --git a/ydb/library/workload/tpch/tpch_schema.yaml b/ydb/library/workload/tpch/tpch_schema.yaml
index 4041d510183..2bcfd8cdcb3 100644
--- a/ydb/library/workload/tpch/tpch_schema.yaml
+++ b/ydb/library/workload/tpch/tpch_schema.yaml
@@ -81,6 +81,8 @@ tables:
     primary_key:
       - l_orderkey
       - l_linenumber
+    partition_by:
+      - l_orderkey
 
   - name: nation
     columns:
diff --git a/ydb/tests/functional/benchmarks_init/canondata/result.json b/ydb/tests/functional/benchmarks_init/canondata/result.json
index 4ae82132a23..ff54afb3c35 100644
--- a/ydb/tests/functional/benchmarks_init/canondata/result.json
+++ b/ydb/tests/functional/benchmarks_init/canondata/result.json
@@ -32,6 +32,9 @@
     "test_init.TestClickbenchInit.test_s1_s3": {
         "uri": "file://test_init.TestClickbenchInit.test_s1_s3/s1_s3"
     },
+    "test_init.TestTpcdsInit.test_s100_column": {
+        "uri": "file://test_init.TestTpcdsInit.test_s100_column/s100_column"
+    },
     "test_init.TestTpcdsInit.test_s1_column": {
         "uri": "file://test_init.TestTpcdsInit.test_s1_column/s1_column"
     },
@@ -47,6 +50,9 @@
     "test_init.TestTpcdsInit.test_s1_s3": {
         "uri": "file://test_init.TestTpcdsInit.test_s1_s3/s1_s3"
     },
+    "test_init.TestTpchInit.test_s100_column": {
+        "uri": "file://test_init.TestTpchInit.test_s100_column/s100_column"
+    },
     "test_init.TestTpchInit.test_s1_column": {
         "uri": "file://test_init.TestTpchInit.test_s1_column/s1_column"
     },
diff --git a/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpcdsInit.test_s100_column/s100_column b/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpcdsInit.test_s100_column/s100_column
new file mode 100644
index 00000000000..158ec1509f1
--- /dev/null
+++ b/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpcdsInit.test_s100_column/s100_column
@@ -0,0 +1,645 @@
+Init tables ...
+--!syntax_v1
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/customer_address` (
+    ca_address_sk Int64 NOT NULL,
+    ca_address_id Utf8,
+    ca_street_number Utf8,
+    ca_street_name Utf8,
+    ca_street_type Utf8,
+    ca_suite_number Utf8,
+    ca_city Utf8,
+    ca_county Utf8,
+    ca_state Utf8,
+    ca_zip Utf8,
+    ca_country Utf8,
+    ca_gmt_offset Double,
+    ca_location_type Utf8,
+    PRIMARY KEY (ca_address_sk)
+)
+PARTITION BY HASH (ca_address_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/customer_demographics` (
+    cd_demo_sk Int64 NOT NULL,
+    cd_gender Utf8,
+    cd_marital_status Utf8,
+    cd_education_status Utf8,
+    cd_purchase_estimate Int64,
+    cd_credit_rating Utf8,
+    cd_dep_count Int64,
+    cd_dep_employed_count Int64,
+    cd_dep_college_count Int64,
+    PRIMARY KEY (cd_demo_sk)
+)
+PARTITION BY HASH (cd_demo_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/date_dim` (
+    d_date_sk Int64 NOT NULL,
+    d_date_id Utf8,
+    d_date Utf8,
+    d_month_seq Int64,
+    d_week_seq Int64,
+    d_quarter_seq Int64,
+    d_year Int64,
+    d_dow Int64,
+    d_moy Int64,
+    d_dom Int64,
+    d_qoy Int64,
+    d_fy_year Int64,
+    d_fy_quarter_seq Int64,
+    d_fy_week_seq Int64,
+    d_day_name Utf8,
+    d_quarter_name Utf8,
+    d_holiday Utf8,
+    d_weekend Utf8,
+    d_following_holiday Utf8,
+    d_first_dom Int64,
+    d_last_dom Int64,
+    d_same_day_ly Int64,
+    d_same_day_lq Int64,
+    d_current_day Utf8,
+    d_current_week Utf8,
+    d_current_month Utf8,
+    d_current_quarter Utf8,
+    d_current_year Utf8,
+    PRIMARY KEY (d_date_sk)
+)
+PARTITION BY HASH (d_date_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/warehouse` (
+    w_warehouse_sk Int64 NOT NULL,
+    w_warehouse_id Utf8,
+    w_warehouse_name Utf8,
+    w_warehouse_sq_ft Int64,
+    w_street_number Utf8,
+    w_street_name Utf8,
+    w_street_type Utf8,
+    w_suite_number Utf8,
+    w_city Utf8,
+    w_county Utf8,
+    w_state Utf8,
+    w_zip Utf8,
+    w_country Utf8,
+    w_gmt_offset Double,
+    PRIMARY KEY (w_warehouse_sk)
+)
+PARTITION BY HASH (w_warehouse_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/ship_mode` (
+    sm_ship_mode_sk Int64 NOT NULL,
+    sm_ship_mode_id Utf8,
+    sm_type Utf8,
+    sm_code Utf8,
+    sm_carrier Utf8,
+    sm_contract Utf8,
+    PRIMARY KEY (sm_ship_mode_sk)
+)
+PARTITION BY HASH (sm_ship_mode_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/time_dim` (
+    t_time_sk Int64 NOT NULL,
+    t_time_id Utf8,
+    t_time Int64,
+    t_hour Int64,
+    t_minute Int64,
+    t_second Int64,
+    t_am_pm Utf8,
+    t_shift Utf8,
+    t_sub_shift Utf8,
+    t_meal_time Utf8,
+    PRIMARY KEY (t_time_sk)
+)
+PARTITION BY HASH (t_time_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/reason` (
+    r_reason_sk Int64 NOT NULL,
+    r_reason_id Utf8,
+    r_reason_desc Utf8,
+    PRIMARY KEY (r_reason_sk)
+)
+PARTITION BY HASH (r_reason_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/income_band` (
+    ib_income_band_sk Int64 NOT NULL,
+    ib_lower_bound Int64,
+    ib_upper_bound Int64,
+    PRIMARY KEY (ib_income_band_sk)
+)
+PARTITION BY HASH (ib_income_band_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/item` (
+    i_item_sk Int64 NOT NULL,
+    i_item_id Utf8,
+    i_rec_start_date Date32,
+    i_rec_end_date Date32,
+    i_item_desc Utf8,
+    i_current_price Double,
+    i_wholesale_cost Double,
+    i_brand_id Int64,
+    i_brand Utf8,
+    i_class_id Int64,
+    i_class Utf8,
+    i_category_id Int64,
+    i_category Utf8,
+    i_manufact_id Int64,
+    i_manufact Utf8,
+    i_size Utf8,
+    i_formulation Utf8,
+    i_color Utf8,
+    i_units Utf8,
+    i_container Utf8,
+    i_manager_id Int64,
+    i_product_name Utf8,
+    PRIMARY KEY (i_item_sk)
+)
+PARTITION BY HASH (i_item_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/store` (
+    s_store_sk Int64 NOT NULL,
+    s_store_id Utf8,
+    s_rec_start_date Date32,
+    s_rec_end_date Date32,
+    s_closed_date_sk Int64,
+    s_store_name Utf8,
+    s_number_employees Int64,
+    s_floor_space Int64,
+    s_hours Utf8,
+    s_manager Utf8,
+    s_market_id Int64,
+    s_geography_class Utf8,
+    s_market_desc Utf8,
+    s_market_manager Utf8,
+    s_division_id Int64,
+    s_division_name Utf8,
+    s_company_id Int64,
+    s_company_name Utf8,
+    s_street_number Utf8,
+    s_street_name Utf8,
+    s_street_type Utf8,
+    s_suite_number Utf8,
+    s_city Utf8,
+    s_county Utf8,
+    s_state Utf8,
+    s_zip Utf8,
+    s_country Utf8,
+    s_gmt_offset Double,
+    s_tax_precentage Double,
+    PRIMARY KEY (s_store_sk)
+)
+PARTITION BY HASH (s_store_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/call_center` (
+    cc_call_center_sk Int64 NOT NULL,
+    cc_call_center_id Utf8,
+    cc_rec_start_date Date32,
+    cc_rec_end_date Date32,
+    cc_closed_date_sk Int64,
+    cc_open_date_sk Int64,
+    cc_name Utf8,
+    cc_class Utf8,
+    cc_employees Int64,
+    cc_sq_ft Int64,
+    cc_hours Utf8,
+    cc_manager Utf8,
+    cc_mkt_id Int64,
+    cc_mkt_class Utf8,
+    cc_mkt_desc Utf8,
+    cc_market_manager Utf8,
+    cc_division Int64,
+    cc_division_name Utf8,
+    cc_company Int64,
+    cc_company_name Utf8,
+    cc_street_number Utf8,
+    cc_street_name Utf8,
+    cc_street_type Utf8,
+    cc_suite_number Utf8,
+    cc_city Utf8,
+    cc_county Utf8,
+    cc_state Utf8,
+    cc_zip Utf8,
+    cc_country Utf8,
+    cc_gmt_offset Double,
+    cc_tax_percentage Double,
+    PRIMARY KEY (cc_call_center_sk)
+)
+PARTITION BY HASH (cc_call_center_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/customer` (
+    c_customer_sk Int64 NOT NULL,
+    c_customer_id Utf8,
+    c_current_cdemo_sk Int64,
+    c_current_hdemo_sk Int64,
+    c_current_addr_sk Int64,
+    c_first_shipto_date_sk Int64,
+    c_first_sales_date_sk Int64,
+    c_salutation Utf8,
+    c_first_name Utf8,
+    c_last_name Utf8,
+    c_preferred_cust_flag Utf8,
+    c_birth_day Int64,
+    c_birth_month Int64,
+    c_birth_year Int64,
+    c_birth_country Utf8,
+    c_login Utf8,
+    c_email_address Utf8,
+    c_last_review_date Utf8,
+    PRIMARY KEY (c_customer_sk)
+)
+PARTITION BY HASH (c_customer_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/web_site` (
+    web_site_sk Int64 NOT NULL,
+    web_site_id Utf8,
+    web_rec_start_date Date32,
+    web_rec_end_date Date32,
+    web_name Utf8,
+    web_open_date_sk Int64,
+    web_close_date_sk Int64,
+    web_class Utf8,
+    web_manager Utf8,
+    web_mkt_id Int64,
+    web_mkt_class Utf8,
+    web_mkt_desc Utf8,
+    web_market_manager Utf8,
+    web_company_id Int64,
+    web_company_name Utf8,
+    web_street_number Utf8,
+    web_street_name Utf8,
+    web_street_type Utf8,
+    web_suite_number Utf8,
+    web_city Utf8,
+    web_county Utf8,
+    web_state Utf8,
+    web_zip Utf8,
+    web_country Utf8,
+    web_gmt_offset Double,
+    web_tax_percentage Double,
+    PRIMARY KEY (web_site_sk)
+)
+PARTITION BY HASH (web_site_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/store_returns` (
+    sr_returned_date_sk Int64,
+    sr_return_time_sk Int64,
+    sr_item_sk Int64 NOT NULL,
+    sr_customer_sk Int64,
+    sr_cdemo_sk Int64,
+    sr_hdemo_sk Int64,
+    sr_addr_sk Int64,
+    sr_store_sk Int64,
+    sr_reason_sk Int64,
+    sr_ticket_number Int64 NOT NULL,
+    sr_return_quantity Int64,
+    sr_return_amt Double,
+    sr_return_tax Double,
+    sr_return_amt_inc_tax Double,
+    sr_fee Double,
+    sr_return_ship_cost Double,
+    sr_refunded_cash Double,
+    sr_reversed_charge Double,
+    sr_store_credit Double,
+    sr_net_loss Double,
+    PRIMARY KEY (sr_item_sk, sr_ticket_number)
+)
+PARTITION BY HASH (sr_item_sk, sr_ticket_number)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/household_demographics` (
+    hd_demo_sk Int64 NOT NULL,
+    hd_income_band_sk Int64,
+    hd_buy_potential Utf8,
+    hd_dep_count Int64,
+    hd_vehicle_count Int64,
+    PRIMARY KEY (hd_demo_sk)
+)
+PARTITION BY HASH (hd_demo_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/web_page` (
+    wp_web_page_sk Int64 NOT NULL,
+    wp_web_page_id Utf8,
+    wp_rec_start_date Date32,
+    wp_rec_end_date Date32,
+    wp_creation_date_sk Int64,
+    wp_access_date_sk Int64,
+    wp_autogen_flag Utf8,
+    wp_customer_sk Int64,
+    wp_url Utf8,
+    wp_type Utf8,
+    wp_char_count Int64,
+    wp_link_count Int64,
+    wp_image_count Int64,
+    wp_max_ad_count Int64,
+    PRIMARY KEY (wp_web_page_sk)
+)
+PARTITION BY HASH (wp_web_page_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/promotion` (
+    p_promo_sk Int64 NOT NULL,
+    p_promo_id Utf8,
+    p_start_date_sk Int64,
+    p_end_date_sk Int64,
+    p_item_sk Int64,
+    p_cost Double,
+    p_response_target Int64,
+    p_promo_name Utf8,
+    p_channel_dmail Utf8,
+    p_channel_email Utf8,
+    p_channel_catalog Utf8,
+    p_channel_tv Utf8,
+    p_channel_radio Utf8,
+    p_channel_press Utf8,
+    p_channel_event Utf8,
+    p_channel_demo Utf8,
+    p_channel_details Utf8,
+    p_purpose Utf8,
+    p_discount_active Utf8,
+    PRIMARY KEY (p_promo_sk)
+)
+PARTITION BY HASH (p_promo_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/catalog_page` (
+    cp_catalog_page_sk Int64 NOT NULL,
+    cp_catalog_page_id Utf8,
+    cp_start_date_sk Int64,
+    cp_end_date_sk Int64,
+    cp_department Utf8,
+    cp_catalog_number Int64,
+    cp_catalog_page_number Int64,
+    cp_description Utf8,
+    cp_type Utf8,
+    PRIMARY KEY (cp_catalog_page_sk)
+)
+PARTITION BY HASH (cp_catalog_page_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/inventory` (
+    inv_date_sk Int64 NOT NULL,
+    inv_item_sk Int64 NOT NULL,
+    inv_warehouse_sk Int64 NOT NULL,
+    inv_quantity_on_hand Int64,
+    PRIMARY KEY (inv_date_sk, inv_item_sk, inv_warehouse_sk)
+)
+PARTITION BY HASH (inv_date_sk, inv_item_sk, inv_warehouse_sk)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/catalog_returns` (
+    cr_returned_date_sk Int64,
+    cr_returned_time_sk Int64,
+    cr_item_sk Int64 NOT NULL,
+    cr_refunded_customer_sk Int64,
+    cr_refunded_cdemo_sk Int64,
+    cr_refunded_hdemo_sk Int64,
+    cr_refunded_addr_sk Int64,
+    cr_returning_customer_sk Int64,
+    cr_returning_cdemo_sk Int64,
+    cr_returning_hdemo_sk Int64,
+    cr_returning_addr_sk Int64,
+    cr_call_center_sk Int64,
+    cr_catalog_page_sk Int64,
+    cr_ship_mode_sk Int64,
+    cr_warehouse_sk Int64,
+    cr_reason_sk Int64,
+    cr_order_number Int64 NOT NULL,
+    cr_return_quantity Int64,
+    cr_return_amount Double,
+    cr_return_tax Double,
+    cr_return_amt_inc_tax Double,
+    cr_fee Double,
+    cr_return_ship_cost Double,
+    cr_refunded_cash Double,
+    cr_reversed_charge Double,
+    cr_store_credit Double,
+    cr_net_loss Double,
+    PRIMARY KEY (cr_item_sk, cr_order_number)
+)
+PARTITION BY HASH (cr_item_sk, cr_order_number)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/web_returns` (
+    wr_returned_date_sk Int64,
+    wr_returned_time_sk Int64,
+    wr_item_sk Int64 NOT NULL,
+    wr_refunded_customer_sk Int64,
+    wr_refunded_cdemo_sk Int64,
+    wr_refunded_hdemo_sk Int64,
+    wr_refunded_addr_sk Int64,
+    wr_returning_customer_sk Int64,
+    wr_returning_cdemo_sk Int64,
+    wr_returning_hdemo_sk Int64,
+    wr_returning_addr_sk Int64,
+    wr_web_page_sk Int64,
+    wr_reason_sk Int64,
+    wr_order_number Int64 NOT NULL,
+    wr_return_quantity Int64,
+    wr_return_amt Double,
+    wr_return_tax Double,
+    wr_return_amt_inc_tax Double,
+    wr_fee Double,
+    wr_return_ship_cost Double,
+    wr_refunded_cash Double,
+    wr_reversed_charge Double,
+    wr_account_credit Double,
+    wr_net_loss Double,
+    PRIMARY KEY (wr_item_sk, wr_order_number)
+)
+PARTITION BY HASH (wr_item_sk, wr_order_number)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/web_sales` (
+    ws_sold_date_sk Int64,
+    ws_sold_time_sk Int64,
+    ws_ship_date_sk Int64,
+    ws_item_sk Int64 NOT NULL,
+    ws_bill_customer_sk Int64,
+    ws_bill_cdemo_sk Int64,
+    ws_bill_hdemo_sk Int64,
+    ws_bill_addr_sk Int64,
+    ws_ship_customer_sk Int64,
+    ws_ship_cdemo_sk Int64,
+    ws_ship_hdemo_sk Int64,
+    ws_ship_addr_sk Int64,
+    ws_web_page_sk Int64,
+    ws_web_site_sk Int64,
+    ws_ship_mode_sk Int64,
+    ws_warehouse_sk Int64,
+    ws_promo_sk Int64,
+    ws_order_number Int64 NOT NULL,
+    ws_quantity Int64,
+    ws_wholesale_cost Double,
+    ws_list_price Double,
+    ws_sales_price Double,
+    ws_ext_discount_amt Double,
+    ws_ext_sales_price Double,
+    ws_ext_wholesale_cost Double,
+    ws_ext_list_price Double,
+    ws_ext_tax Double,
+    ws_coupon_amt Double,
+    ws_ext_ship_cost Double,
+    ws_net_paid Double,
+    ws_net_paid_inc_tax Double,
+    ws_net_paid_inc_ship Double,
+    ws_net_paid_inc_ship_tax Double,
+    ws_net_profit Double,
+    PRIMARY KEY (ws_item_sk, ws_order_number)
+)
+PARTITION BY HASH (ws_item_sk, ws_order_number)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/catalog_sales` (
+    cs_sold_date_sk Int64,
+    cs_sold_time_sk Int64,
+    cs_ship_date_sk Int64,
+    cs_bill_customer_sk Int64,
+    cs_bill_cdemo_sk Int64,
+    cs_bill_hdemo_sk Int64,
+    cs_bill_addr_sk Int64,
+    cs_ship_customer_sk Int64,
+    cs_ship_cdemo_sk Int64,
+    cs_ship_hdemo_sk Int64,
+    cs_ship_addr_sk Int64,
+    cs_call_center_sk Int64,
+    cs_catalog_page_sk Int64,
+    cs_ship_mode_sk Int64,
+    cs_warehouse_sk Int64,
+    cs_item_sk Int64 NOT NULL,
+    cs_promo_sk Int64,
+    cs_order_number Int64 NOT NULL,
+    cs_quantity Int64,
+    cs_wholesale_cost Double,
+    cs_list_price Double,
+    cs_sales_price Double,
+    cs_ext_discount_amt Double,
+    cs_ext_sales_price Double,
+    cs_ext_wholesale_cost Double,
+    cs_ext_list_price Double,
+    cs_ext_tax Double,
+    cs_coupon_amt Double,
+    cs_ext_ship_cost Double,
+    cs_net_paid Double,
+    cs_net_paid_inc_tax Double,
+    cs_net_paid_inc_ship Double,
+    cs_net_paid_inc_ship_tax Double,
+    cs_net_profit Double,
+    PRIMARY KEY (cs_item_sk, cs_order_number)
+)
+PARTITION BY HASH (cs_item_sk, cs_order_number)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+CREATE TABLE `/Root/db/Root/db/tpcds/s1/store_sales` (
+    ss_sold_date_sk Int64,
+    ss_sold_time_sk Int64,
+    ss_item_sk Int64 NOT NULL,
+    ss_customer_sk Int64,
+    ss_cdemo_sk Int64,
+    ss_hdemo_sk Int64,
+    ss_addr_sk Int64,
+    ss_store_sk Int64,
+    ss_promo_sk Int64,
+    ss_ticket_number Int64 NOT NULL,
+    ss_quantity Int64,
+    ss_wholesale_cost Double,
+    ss_list_price Double,
+    ss_sales_price Double,
+    ss_ext_discount_amt Double,
+    ss_ext_sales_price Double,
+    ss_ext_wholesale_cost Double,
+    ss_ext_list_price Double,
+    ss_ext_tax Double,
+    ss_coupon_amt Double,
+    ss_net_paid Double,
+    ss_net_paid_inc_tax Double,
+    ss_net_profit Double,
+    PRIMARY KEY (ss_item_sk, ss_ticket_number)
+)
+PARTITION BY HASH (ss_item_sk, ss_ticket_number)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
+);
+
+Init tables ...Ok
diff --git a/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpchInit.test_s100_column/s100_column b/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpchInit.test_s100_column/s100_column
new file mode 100644
index 00000000000..47806ca50fa
--- /dev/null
+++ b/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpchInit.test_s100_column/s100_column
@@ -0,0 +1,137 @@
+Init tables ...
+--!syntax_v1
+
+CREATE TABLE `/Root/db/Root/db/tpch/s1/customer` (
+    c_acctbal Double NOT NULL,
+    c_address Utf8 NOT NULL,
+    c_comment Utf8 NOT NULL,
+    c_custkey Int64 NOT NULL,
+    c_mktsegment Utf8 NOT NULL,
+    c_name Utf8 NOT NULL,
+    c_nationkey Int32 NOT NULL,
+    c_phone Utf8 NOT NULL,
+    PRIMARY KEY (c_custkey)
+)
+PARTITION BY HASH (c_custkey)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 256
+);
+
+CREATE TABLE `/Root/db/Root/db/tpch/s1/lineitem` (
+    l_comment Utf8 NOT NULL,
+    l_commitdate Date32 NOT NULL,
+    l_discount Double NOT NULL,
+    l_extendedprice Double NOT NULL,
+    l_linenumber Int32 NOT NULL,
+    l_linestatus Utf8 NOT NULL,
+    l_orderkey Int64 NOT NULL,
+    l_partkey Int64 NOT NULL,
+    l_quantity Double NOT NULL,
+    l_receiptdate Date32 NOT NULL,
+    l_returnflag Utf8 NOT NULL,
+    l_shipdate Date32 NOT NULL,
+    l_shipinstruct Utf8 NOT NULL,
+    l_shipmode Utf8 NOT NULL,
+    l_suppkey Int64 NOT NULL,
+    l_tax Double NOT NULL,
+    PRIMARY KEY (l_orderkey, l_linenumber)
+)
+PARTITION BY HASH (l_orderkey)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 256
+);
+
+CREATE TABLE `/Root/db/Root/db/tpch/s1/nation` (
+    n_comment Utf8 NOT NULL,
+    n_name Utf8 NOT NULL,
+    n_nationkey Int32 NOT NULL,
+    n_regionkey Int32 NOT NULL,
+    PRIMARY KEY (n_nationkey)
+)
+PARTITION BY HASH (n_nationkey)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 1
+);
+
+CREATE TABLE `/Root/db/Root/db/tpch/s1/orders` (
+    o_clerk Utf8 NOT NULL,
+    o_comment Utf8 NOT NULL,
+    o_custkey Int64 NOT NULL,
+    o_orderdate Date32 NOT NULL,
+    o_orderkey Int64 NOT NULL,
+    o_orderpriority Utf8 NOT NULL,
+    o_orderstatus Utf8 NOT NULL,
+    o_shippriority Int32 NOT NULL,
+    o_totalprice Double NOT NULL,
+    PRIMARY KEY (o_orderkey)
+)
+PARTITION BY HASH (o_orderkey)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 256
+);
+
+CREATE TABLE `/Root/db/Root/db/tpch/s1/part` (
+    p_brand Utf8 NOT NULL,
+    p_comment Utf8 NOT NULL,
+    p_container Utf8 NOT NULL,
+    p_mfgr Utf8 NOT NULL,
+    p_name Utf8 NOT NULL,
+    p_partkey Int64 NOT NULL,
+    p_retailprice Double NOT NULL,
+    p_size Int32 NOT NULL,
+    p_type Utf8 NOT NULL,
+    PRIMARY KEY (p_partkey)
+)
+PARTITION BY HASH (p_partkey)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 256
+);
+
+CREATE TABLE `/Root/db/Root/db/tpch/s1/partsupp` (
+    ps_availqty Int32 NOT NULL,
+    ps_comment Utf8 NOT NULL,
+    ps_partkey Int64 NOT NULL,
+    ps_suppkey Int64 NOT NULL,
+    ps_supplycost Double NOT NULL,
+    PRIMARY KEY (ps_partkey, ps_suppkey)
+)
+PARTITION BY HASH (ps_partkey, ps_suppkey)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 256
+);
+
+CREATE TABLE `/Root/db/Root/db/tpch/s1/region` (
+    r_comment Utf8 NOT NULL,
+    r_name Utf8 NOT NULL,
+    r_regionkey Int32 NOT NULL,
+    PRIMARY KEY (r_regionkey)
+)
+PARTITION BY HASH (r_regionkey)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 1
+);
+
+CREATE TABLE `/Root/db/Root/db/tpch/s1/supplier` (
+    s_acctbal Double NOT NULL,
+    s_address Utf8 NOT NULL,
+    s_comment Utf8 NOT NULL,
+    s_name Utf8 NOT NULL,
+    s_nationkey Int32 NOT NULL,
+    s_phone Utf8 NOT NULL,
+    s_suppkey Int64 NOT NULL,
+    PRIMARY KEY (s_suppkey)
+)
+PARTITION BY HASH (s_suppkey)
+WITH (
+    STORE = COLUMN,
+    AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 256
+);
+
+Init tables ...Ok
diff --git a/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpchInit.test_s1_column/s1_column b/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpchInit.test_s1_column/s1_column
index 45e7e60ce35..098d7418a43 100644
--- a/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpchInit.test_s1_column/s1_column
+++ b/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpchInit.test_s1_column/s1_column
@@ -37,7 +37,7 @@ CREATE TABLE `/Root/db/Root/db/tpch/s1/lineitem` (
     l_tax Double NOT NULL,
     PRIMARY KEY (l_orderkey, l_linenumber)
 )
-PARTITION BY HASH (l_orderkey, l_linenumber)
+PARTITION BY HASH (l_orderkey)
 WITH (
     STORE = COLUMN,
     AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
diff --git a/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpchInit.test_s1_column_decimal/s1_column_decimal b/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpchInit.test_s1_column_decimal/s1_column_decimal
index a87903fc168..fe6e09bd1a7 100644
--- a/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpchInit.test_s1_column_decimal/s1_column_decimal
+++ b/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpchInit.test_s1_column_decimal/s1_column_decimal
@@ -37,7 +37,7 @@ CREATE TABLE `/Root/db/Root/db/tpch/s1/lineitem` (
     l_tax Decimal(12,2) NOT NULL,
     PRIMARY KEY (l_orderkey, l_linenumber)
 )
-PARTITION BY HASH (l_orderkey, l_linenumber)
+PARTITION BY HASH (l_orderkey)
 WITH (
     STORE = COLUMN,
     AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
diff --git a/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpchInit.test_s1_column_decimal_ydb/s1_column_decimal_ydb b/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpchInit.test_s1_column_decimal_ydb/s1_column_decimal_ydb
index 2ceafd6902d..a29220d4795 100644
--- a/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpchInit.test_s1_column_decimal_ydb/s1_column_decimal_ydb
+++ b/ydb/tests/functional/benchmarks_init/canondata/test_init.TestTpchInit.test_s1_column_decimal_ydb/s1_column_decimal_ydb
@@ -37,7 +37,7 @@ CREATE TABLE `/Root/db/Root/db/tpch/s1/lineitem` (
     l_tax Decimal(22,9) NOT NULL,
     PRIMARY KEY (l_orderkey, l_linenumber)
 )
-PARTITION BY HASH (l_orderkey, l_linenumber)
+PARTITION BY HASH (l_orderkey)
 WITH (
     STORE = COLUMN,
     AUTO_PARTITIONING_MIN_PARTITIONS_COUNT = 64
diff --git a/ydb/tests/functional/benchmarks_init/test_init.py b/ydb/tests/functional/benchmarks_init/test_init.py
index f20511ead93..5f9137e4da8 100644
--- a/ydb/tests/functional/benchmarks_init/test_init.py
+++ b/ydb/tests/functional/benchmarks_init/test_init.py
@@ -79,6 +79,9 @@ class TpcInitBase(InitBase):
     def test_s1_column_decimal_ydb(self):
         return self.canonical_result(self.execute_init(scale=1, args=['--store', 'column', '--float-mode', 'decimal_ydb']), self.tmp_path('s1_column_decimal_ydb'))
 
+    def test_s100_column(self):
+        return self.canonical_result(self.execute_init(scale=1, args=['--store', 'column', '--scale', '100']), self.tmp_path('s100_column'))
+
 
 class TestTpchInit(TpcInitBase):
     workload = 'tpch'