diff --git a/be/src/exec/json_scanner.cpp b/be/src/exec/json_scanner.cpp
index 6143a9e19c..ca6cd5622c 100644
--- a/be/src/exec/json_scanner.cpp
+++ b/be/src/exec/json_scanner.cpp
@@ -134,16 +134,20 @@ Status JsonScanner::open_next_reader() {
     }
     }
 
+    std::string json_root = "";
     std::string jsonpath = "";
     bool strip_outer_array = false;
     if (range.__isset.jsonpaths) {
         jsonpath = range.jsonpaths;
     }
+    if (range.__isset.json_root) {
+        json_root = range.json_root;
+    }
     if (range.__isset.strip_outer_array) {
         strip_outer_array = range.strip_outer_array;
     }
-    _cur_file_reader = new JsonReader(_state, _counter, _profile, file, jsonpath, strip_outer_array);
-    RETURN_IF_ERROR(_cur_file_reader->init());
+    _cur_file_reader = new JsonReader(_state, _counter, _profile, file, strip_outer_array);
+    RETURN_IF_ERROR(_cur_file_reader->init(jsonpath, json_root));
 
     return Status::OK();
 }
@@ -178,9 +182,8 @@ JsonReader::JsonReader(
         RuntimeState* state, ScannerCounter* counter,
         RuntimeProfile* profile,
         FileReader* file_reader,
-        std::string& jsonpath,
         bool strip_outer_array) :
-            _jsonpath(jsonpath),
+            _handle_json_callback(nullptr),
             _next_line(0),
             _total_lines(0),
             _state(state),
@@ -188,7 +191,8 @@ JsonReader::JsonReader(
             _profile(profile),
             _file_reader(file_reader),
             _closed(false),
-            _strip_outer_array(strip_outer_array) {
+            _strip_outer_array(strip_outer_array),
+            _json_doc(nullptr) {
     _bytes_read_counter = ADD_COUNTER(_profile, "BytesRead", TUnit::BYTES);
     _read_timer = ADD_TIMER(_profile, "FileReadTime");
 }
@@ -197,32 +201,52 @@ JsonReader::~JsonReader() {
     _close();
 }
 
-Status JsonReader::init() {
+Status JsonReader::init(const std::string& jsonpath, const std::string& json_root) {
     // parse jsonpath
-    rapidjson::Document jsonpaths_doc;
-    if (!_jsonpath.empty()) {
-        if (!jsonpaths_doc.Parse(_jsonpath.c_str()).HasParseError()) {
-            if (!jsonpaths_doc.IsArray()) {
-                return Status::InvalidArgument("Invalid json path: " + _jsonpath);
-            } else {
-                for (int i = 0; i < jsonpaths_doc.Size(); i++) {
-                    const rapidjson::Value& path = jsonpaths_doc[i];
-                    if (!path.IsString()) {
-                        return Status::InvalidArgument("Invalid json path: " + _jsonpath);
-                    }
-                    path.GetString();    
-                    std::vector<JsonPath> parsed_paths;
-                    JsonFunctions::parse_json_paths(path.GetString(), &parsed_paths);
-                    _parsed_jsonpaths.push_back(parsed_paths);
-                }
-            }
+    if (!jsonpath.empty()) {
+        Status st = _generate_json_paths(jsonpath, &_parsed_jsonpaths);
+        RETURN_IF_ERROR(st);
+    }
+    if (!json_root.empty()) {
+        JsonFunctions::parse_json_paths(json_root, &_parsed_json_root);
+    }
+
+    //improve performance
+    if (_parsed_jsonpaths.empty()) { // input is a simple json-string
+        _handle_json_callback = &JsonReader::_handle_simple_json;
+    } else { // input is a complex json-string and a json-path
+        if (_strip_outer_array) {
+            _handle_json_callback = &JsonReader::_handle_flat_array_complex_json;
         } else {
-            return Status::InvalidArgument("Invalid json path: " + _jsonpath);
+            _handle_json_callback = &JsonReader::_handle_nested_complex_json;
         }
     }
     return Status::OK();
 }
 
+Status JsonReader::_generate_json_paths(const std::string& jsonpath, std::vector<std::vector<JsonPath>>* vect) {
+    rapidjson::Document jsonpaths_doc;
+    if (!jsonpaths_doc.Parse(jsonpath.c_str()).HasParseError()) {
+        if (!jsonpaths_doc.IsArray()) {
+            return Status::InvalidArgument("Invalid json path: " + jsonpath);
+        } else {
+            for (int i = 0; i < jsonpaths_doc.Size(); i++) {
+                const rapidjson::Value& path = jsonpaths_doc[i];
+                if (!path.IsString()) {
+                    return Status::InvalidArgument("Invalid json path: " + jsonpath);
+                }
+                path.GetString();
+                std::vector<JsonPath> parsed_paths;
+                JsonFunctions::parse_json_paths(path.GetString(), &parsed_paths);
+                vect->push_back(parsed_paths);
+            }
+            return Status::OK();
+        }
+    } else {
+        return Status::InvalidArgument("Invalid json path: " + jsonpath);
+    }
+}
+
 void JsonReader::_close() {
     if (_closed) {
         return;
@@ -248,35 +272,47 @@ Status JsonReader::_parse_json_doc(bool* eof) {
         return Status::OK();
     }
     // parse jsondata to JsonDoc
-    if (_json_doc.Parse((char*)json_str, length).HasParseError()) {
+    if (_origin_json_doc.Parse((char*)json_str, length).HasParseError()) {
         std::stringstream str_error;
-        str_error << "Parse json data for JsonDoc failed. code = " << _json_doc.GetParseError()
-                << ", error-info:" << rapidjson::GetParseError_En(_json_doc.GetParseError());
+        str_error << "Parse json data for JsonDoc failed. code = " << _origin_json_doc.GetParseError()
+                << ", error-info:" << rapidjson::GetParseError_En(_origin_json_doc.GetParseError());
         _state->append_error_msg_to_file(std::string((char*) json_str, length), str_error.str());
         _counter->num_rows_filtered++;
         delete[] json_str;
         return Status::DataQualityError(str_error.str());
     }
+    delete[] json_str;
 
-    if (_json_doc.IsArray() && !_strip_outer_array) {
-        delete[] json_str;
+    // set json root
+    if (_parsed_json_root.size() != 0) {
+        _json_doc = JsonFunctions::get_json_object_from_parsed_json(_parsed_json_root, &_origin_json_doc, _origin_json_doc.GetAllocator());
+        if (_json_doc == nullptr) {
+            std::stringstream str_error;
+            str_error << "JSON Root not found.";
+            _state->append_error_msg_to_file(_print_json_value(_origin_json_doc), str_error.str());
+            _counter->num_rows_filtered++;
+            return Status::DataQualityError(str_error.str());
+        }
+    } else {
+        _json_doc = &_origin_json_doc;
+    }
+
+    if (_json_doc->IsArray() && !_strip_outer_array) {
         std::stringstream str_error;
         str_error << "JSON data is array-object, `strip_outer_array` must be TRUE.";
-        _state->append_error_msg_to_file(_print_json_value(_json_doc), str_error.str());
+        _state->append_error_msg_to_file(_print_json_value(_origin_json_doc), str_error.str());
         _counter->num_rows_filtered++;
         return Status::DataQualityError(str_error.str());
     }
 
-    if (!_json_doc.IsArray() && _strip_outer_array) {
-        delete[] json_str;
+    if (!_json_doc->IsArray() && _strip_outer_array) {
         std::stringstream str_error;
         str_error << "JSON data is not an array-object, `strip_outer_array` must be FALSE.";
-        _state->append_error_msg_to_file(_print_json_value(_json_doc), str_error.str());
+        _state->append_error_msg_to_file(_print_json_value(_origin_json_doc), str_error.str());
         _counter->num_rows_filtered++;
         return Status::DataQualityError(str_error.str());
     }
 
-    delete[] json_str;
     return Status::OK();
 }
 
@@ -288,27 +324,6 @@ std::string JsonReader::_print_json_value(const rapidjson::Value& value) {
     return std::string(buffer.GetString());
 }
 
-void JsonReader::_assemble_jmap(
-        const std::vector<SlotDescriptor*>& slot_descs) {
-    // iterator jsonpath to find object and save it to Map
-    _jmap.clear();
-
-    for (int i = 0; i < _parsed_jsonpaths.size(); i++) {
-        rapidjson::Value* json_value = JsonFunctions::get_json_array_from_parsed_json(_parsed_jsonpaths[i], &_json_doc, _json_doc.GetAllocator());
-        if (json_value == nullptr) {
-            // null means failed to match, we also put it in _jmap, it will be handled later.
-            _jmap.emplace(slot_descs[i]->col_name(), nullptr);
-        } else {
-            if (json_value->Size() == 1) {
-                // see NOTICE1
-                json_value = &((*json_value)[0]);
-            }
-            _jmap.emplace(slot_descs[i]->col_name(), json_value);
-        }
-    }
-    return;
-}
-
 std::string JsonReader::_print_jsonpath(const std::vector<JsonPath>& path) {
     std::stringstream ss;
     for (auto& p : path) {
@@ -421,7 +436,8 @@ void JsonReader::_set_tuple_value(rapidjson::Value& objectValue, Tuple* tuple, c
  * handle input a simple json.
  * A json is a simple json only when user not specifying the json path.
  * For example:
- *  case 1. [{"colunm1":"value1", "colunm2":10}, {"colunm1":"value2", "colunm2":30}]
+ *  case 1. [{"colunm1":"value1", "colunm2":10}, {"colunm1":"
+", "colunm2":30}]
  *  case 2. {"colunm1":"value1", "colunm2":10}
  */
 Status JsonReader::_handle_simple_json(Tuple* tuple, const std::vector<SlotDescriptor*>& slot_descs, MemPool* tuple_pool, bool* eof) {
@@ -436,19 +452,19 @@ Status JsonReader::_handle_simple_json(Tuple* tuple, const std::vector<SlotDescr
             if (*eof) {// read all data, then return
                 return Status::OK();
             }
-            if (_json_doc.IsArray() ) {
-                _total_lines = _json_doc.Size();
+            if (_json_doc->IsArray()) {
+                _total_lines = _json_doc->Size();
             } else {
                 _total_lines = 1; // only one row
             }
             _next_line = 0;
         }
 
-        if (_json_doc.IsArray()) { // handle case 1
-            rapidjson::Value& objectValue = _json_doc[_next_line];// json object
+        if (_json_doc->IsArray()) { // handle case 1
+            rapidjson::Value& objectValue = (*_json_doc)[_next_line];// json object
             _set_tuple_value(objectValue, tuple, slot_descs, tuple_pool, &valid);
         } else { // handle case 2
-            _set_tuple_value(_json_doc, tuple, slot_descs, tuple_pool, &valid);
+            _set_tuple_value(*_json_doc, tuple, slot_descs, tuple_pool, &valid);
         }
         _next_line++;
         if (!valid) {
@@ -459,64 +475,68 @@ Status JsonReader::_handle_simple_json(Tuple* tuple, const std::vector<SlotDescr
     return Status::OK();
 }
 
-// for complex format json with strip_outer_array = false
-Status JsonReader::_set_tuple_value_from_jmap(Tuple* tuple, const std::vector<SlotDescriptor*>& slot_descs, MemPool* tuple_pool, bool *valid) {
-    std::unordered_map<std::string, JsonDataInternal>::iterator it_map;
-    for (auto v : slot_descs) {
-        it_map = _jmap.find(v->col_name());
-        if (it_map == _jmap.end()) {
-            return Status::RuntimeError("The column name of table is not foud in jsonpath: " + v->col_name());
-        }
-        rapidjson::Value* value = it_map->second.get_value();
-        if (value == nullptr) {
-            if (v->is_nullable()) {
-                tuple->set_null(v->null_indicator_offset());
+bool JsonReader::_write_values_by_jsonpath(rapidjson::Value& objectValue, MemPool* tuple_pool, Tuple* tuple, const std::vector<SlotDescriptor*>& slot_descs) {
+    int nullcount = 0;
+    bool valid = true;
+    size_t column_num = std::min(slot_descs.size(), _parsed_jsonpaths.size());
+
+    for (size_t i = 0; i < column_num; i++) {
+        rapidjson::Value* json_values = JsonFunctions::get_json_array_from_parsed_json(_parsed_jsonpaths[i], &objectValue, _origin_json_doc.GetAllocator());
+        if (json_values == nullptr) {
+            // not match in jsondata.
+            if (slot_descs[i]->is_nullable()) {
+                tuple->set_null(slot_descs[i]->null_indicator_offset());
+                nullcount++;
             } else  {
                 std::stringstream str_error;
-                str_error << "The column `" << it_map->first << "` is not nullable, but it's not found in jsondata.";
-                _state->append_error_msg_to_file(_print_json_value(*value), str_error.str());
+                str_error << "The column `" << slot_descs[i]->col_name() << "` is not nullable, but it's not found in jsondata.";
+                _state->append_error_msg_to_file(_print_json_value(objectValue), str_error.str());
                 _counter->num_rows_filtered++;
-                *valid = false; // current row is invalid
+                valid = false; // current row is invalid
                 break;
             }
         } else {
-            _write_data_to_tuple(value, v, tuple, tuple_pool, valid);
-            if (!(*valid)) {
-                return Status::OK();
+            CHECK(json_values->IsArray());
+            CHECK(json_values->Size() >= 1);
+            if (json_values->Size() == 1) {
+                // NOTICE1: JsonFunctions::get_json_array_from_parsed_json() will wrap the single json object with an array.
+                // so here we unwrap the array to get the real element.
+                // if json_values' size > 1, it means we just match an array, not a wrapped one, so no need to unwrap.
+                json_values = &((*json_values)[0]);
+            }
+            _write_data_to_tuple(json_values, slot_descs[i], tuple, tuple_pool, &valid);
+            if (!valid) {
+                break;
             }
         }
     }
-    *valid = true;
-    return Status::OK();
+    if (nullcount == column_num) {
+        _state->append_error_msg_to_file(_print_json_value(objectValue), "All fields is null or not matched, this is a invalid row.");
+        _counter->num_rows_filtered++;
+    }
+    return valid;
 }
 
-// _json_doc should be an object
+/**
+ * for example:
+ * {
+ *    "data": {"a":"a1", "b":"b1", "c":"c1"}
+ * }
+ * In this scene, generate only one row
+ */
 Status JsonReader::_handle_nested_complex_json(Tuple* tuple, const std::vector<SlotDescriptor*>& slot_descs, MemPool* tuple_pool, bool* eof) {
-    do {
-        bool valid = false;
-        if (_next_line >= _total_lines) {
-            Status st = _parse_json_doc(eof);
-            if (st.is_data_quality_error()) {
-                continue; // continue to read next
-            }
-            RETURN_IF_ERROR(st); // terminate if encounter other errors
-            if (*eof) {
-                return Status::OK();
-            }
-            _assemble_jmap(slot_descs);
-            // for json with strip_outer_array is false, we treat it as a single row.
-            // so _total_lines is always 1.
-            _total_lines = 1;
-            _next_line = 0;
+    while(true) {
+        Status st = _parse_json_doc(eof);
+        if (st.is_data_quality_error()) {
+            continue; // continue to read next
         }
-
-        RETURN_IF_ERROR(_set_tuple_value_from_jmap(tuple, slot_descs, tuple_pool, &valid));
-        _next_line++;
-        if (!valid) { // read a invalid row, then read next one
-            continue;
+        RETURN_IF_ERROR(st);
+        if (*eof) {
+            return Status::OK();// read over,then return
         }
-        break; // read a valid row, then break
-    } while (_next_line <= _total_lines);
+        break; //read a valid row
+    }
+    _write_values_by_jsonpath(*_json_doc, tuple_pool, tuple, slot_descs);
     return Status::OK();
 }
 
@@ -541,73 +561,20 @@ Status JsonReader::_handle_flat_array_complex_json(Tuple* tuple, const std::vect
             if (*eof) {// read all data, then return
                 return Status::OK();
             }
-            _total_lines = _json_doc.Size();
+            _total_lines = _json_doc->Size();
             _next_line = 0;
         }
-        int nullcount = 0;
-        bool valid = true;
-        size_t column_num = std::min(slot_descs.size(), _parsed_jsonpaths.size());
-        rapidjson::Value& objectValue = _json_doc[_next_line];
-
-        for (size_t i = 0; i < column_num; i++) {
-            rapidjson::Value* json_values = JsonFunctions::get_json_array_from_parsed_json(_parsed_jsonpaths[i], &objectValue, _json_doc.GetAllocator());
-            if (json_values == nullptr) {
-                // not match in jsondata.
-                if (slot_descs[i]->is_nullable()) {
-                    tuple->set_null(slot_descs[i]->null_indicator_offset());
-                    nullcount++;
-                } else  {
-                    std::stringstream str_error;
-                    str_error << "The column `" << slot_descs[i]->col_name() << "` is not nullable, but it's not found in jsondata.";
-                    _state->append_error_msg_to_file(_print_json_value(objectValue), str_error.str());
-                    _counter->num_rows_filtered++;
-                    valid = false; // current row is invalid
-                    break;
-                }
-            } else {
-                CHECK(json_values->IsArray());
-                CHECK(json_values->Size() >= 1);
-                if (json_values->Size() == 1) {
-                    // NOTICE1: JsonFunctions::get_json_array_from_parsed_json() will wrap the single json object with an array.
-                    // so here we unwrap the array to get the real element.
-                    // if json_values' size > 1, it means we just match an array, not a wrapped one, so no need to unwrap.
-                    json_values = &((*json_values)[0]);
-                }
-                _write_data_to_tuple(json_values, slot_descs[i], tuple, tuple_pool, &valid);
-                if (!valid) {
-                    break;
-                }
-            }
-        }
-        _next_line++;
-        if (!valid) {
-            continue;
-        }
-        if (nullcount == column_num) {
-            _state->append_error_msg_to_file(_print_json_value(objectValue), "All fields is null or not matched, this is a invalid row.");
-            _counter->num_rows_filtered++;
-            continue;
+        rapidjson::Value& objectValue = (*_json_doc)[_next_line++];
+        if (!_write_values_by_jsonpath(objectValue, tuple_pool, tuple, slot_descs)) {
+            continue; // process next line
         }
         break; // get a valid row, then break
     } while (_next_line <= _total_lines);
     return Status::OK();
 }
 
-// handle json with specified json path
-Status JsonReader::_handle_complex_json(Tuple* tuple, const std::vector<SlotDescriptor*>& slot_descs, MemPool* tuple_pool, bool* eof) {
-    if (_strip_outer_array) {
-        return _handle_flat_array_complex_json(tuple, slot_descs, tuple_pool, eof);
-    } else {
-        return _handle_nested_complex_json(tuple, slot_descs, tuple_pool, eof);
-    }
-}
-
 Status JsonReader::read(Tuple* tuple, const std::vector<SlotDescriptor*>& slot_descs, MemPool* tuple_pool, bool* eof) {
-    if (_parsed_jsonpaths.empty()) { // input is a simple json-string
-        return _handle_simple_json(tuple, slot_descs, tuple_pool, eof);
-    } else { // input is a complex json-string and a json-path
-        return _handle_complex_json(tuple,  slot_descs, tuple_pool, eof);
-    }
+    return (this->*_handle_json_callback)(tuple,  slot_descs, tuple_pool, eof);
 }
 
 
diff --git a/be/src/exec/json_scanner.h b/be/src/exec/json_scanner.h
index fcb1186f31..6dae93159d 100644
--- a/be/src/exec/json_scanner.h
+++ b/be/src/exec/json_scanner.h
@@ -90,7 +90,6 @@ public:
     JsonDataInternal(rapidjson::Value* v);
     ~JsonDataInternal() {}
     rapidjson::Value::ConstValueIterator get_next();
-    rapidjson::Value* get_value() { return _json_values; }
     bool is_null() const { return _json_values == nullptr; }
 
 private:
@@ -106,33 +105,31 @@ struct JsonPath;
 class JsonReader {
 public:
     JsonReader(RuntimeState* state, ScannerCounter* counter, RuntimeProfile* profile, FileReader* file_reader,
-            std::string& jsonpath, bool strip_outer_array);
+            bool strip_outer_array);
     ~JsonReader();
 
-    Status init(); // must call before use
+    Status init(const std::string& jsonpath, const std::string& json_root); // must call before use
 
     Status read(Tuple* tuple, const std::vector<SlotDescriptor*>& slot_descs, MemPool* tuple_pool, bool* eof);
 
 private:
+    Status (JsonReader::*_handle_json_callback)(Tuple* tuple, const std::vector<SlotDescriptor*>& slot_descs, MemPool* tuple_pool, bool* eof);
     Status _handle_simple_json(Tuple* tuple, const std::vector<SlotDescriptor*>& slot_descs, MemPool* tuple_pool, bool* eof);
-    Status _handle_complex_json(Tuple* tuple, const std::vector<SlotDescriptor*>& slot_descs, MemPool* tuple_pool, bool* eof);
     Status _handle_flat_array_complex_json(Tuple* tuple, const std::vector<SlotDescriptor*>& slot_descs, MemPool* tuple_pool, bool* eof);
     Status _handle_nested_complex_json(Tuple* tuple, const std::vector<SlotDescriptor*>& slot_descs, MemPool* tuple_pool, bool* eof);
 
     void _fill_slot(Tuple* tuple, SlotDescriptor* slot_desc, MemPool* mem_pool, const uint8_t* value, int32_t len);
-    void _assemble_jmap(const std::vector<SlotDescriptor*>& slot_descs);
     Status _parse_json_doc(bool* eof);
     void _set_tuple_value(rapidjson::Value& objectValue, Tuple* tuple, const std::vector<SlotDescriptor*>& slot_descs, MemPool* tuple_pool, bool *valid);
-    Status _set_tuple_value_from_jmap(Tuple* tuple, const std::vector<SlotDescriptor*>& slot_descs, MemPool* tuple_pool, bool *valid);
     void _write_data_to_tuple(rapidjson::Value::ConstValueIterator value, SlotDescriptor* desc, Tuple* tuple, MemPool* tuple_pool, bool* valid);
+    bool _write_values_by_jsonpath(rapidjson::Value& objectValue, MemPool* tuple_pool, Tuple* tuple, const std::vector<SlotDescriptor*>& slot_descs);
     std::string _print_json_value(const rapidjson::Value& value);
     std::string _print_jsonpath(const std::vector<JsonPath>& path);
 
     void _close();
+    Status _generate_json_paths(const std::string& jsonpath, std::vector<std::vector<JsonPath>>* vect);
 
 private:
-    // origin json path
-    std::string _jsonpath;
     int _next_line;
     int _total_lines;
     RuntimeState* _state;
@@ -143,10 +140,12 @@ private:
     bool _strip_outer_array;
     RuntimeProfile::Counter* _bytes_read_counter;
     RuntimeProfile::Counter* _read_timer;
+
     std::vector<std::vector<JsonPath>> _parsed_jsonpaths;
-    rapidjson::Document _json_doc;
-    //key: column name
-    std::unordered_map<std::string, JsonDataInternal> _jmap;
+    std::vector<JsonPath> _parsed_json_root;
+
+    rapidjson::Document _origin_json_doc; // origin json document object from parsed json string
+    rapidjson::Value *_json_doc; // _json_doc equals _final_json_doc iff not set `json_root`
 };
 
 
diff --git a/be/src/exprs/json_functions.cpp b/be/src/exprs/json_functions.cpp
index df2510d9e7..8deead4cbe 100644
--- a/be/src/exprs/json_functions.cpp
+++ b/be/src/exprs/json_functions.cpp
@@ -299,6 +299,23 @@ rapidjson::Value* JsonFunctions::get_json_array_from_parsed_json(
     return root;
 }
 
+
+rapidjson::Value* JsonFunctions::get_json_object_from_parsed_json(
+        const std::vector<JsonPath>& parsed_paths,
+        rapidjson::Value* document,
+        rapidjson::Document::AllocatorType& mem_allocator) {
+
+    if (!parsed_paths[0].is_valid) {
+        return nullptr;
+    }
+
+    rapidjson::Value* root = match_value(parsed_paths, document, mem_allocator, true);
+    if (root == nullptr || root == document) { // not found
+        return nullptr;
+    }
+    return root;
+}
+
 void JsonFunctions::json_path_prepare(
         doris_udf::FunctionContext* context,
         doris_udf::FunctionContext::FunctionStateScope scope) {
diff --git a/be/src/exprs/json_functions.h b/be/src/exprs/json_functions.h
index d879e32a5f..657a9c75f3 100644
--- a/be/src/exprs/json_functions.h
+++ b/be/src/exprs/json_functions.h
@@ -95,6 +95,11 @@ public:
             rapidjson::Value* document,
             rapidjson::Document::AllocatorType& mem_allocator);
 
+    static rapidjson::Value* get_json_object_from_parsed_json(
+            const std::vector<JsonPath>& parsed_paths,
+            rapidjson::Value* document,
+            rapidjson::Document::AllocatorType& mem_allocator);
+
     static void json_path_prepare(
             doris_udf::FunctionContext*,
             doris_udf::FunctionContext::FunctionStateScope);
diff --git a/be/src/http/action/stream_load.cpp b/be/src/http/action/stream_load.cpp
index 6e9a616bd8..52e16a5ebc 100644
--- a/be/src/http/action/stream_load.cpp
+++ b/be/src/http/action/stream_load.cpp
@@ -371,11 +371,14 @@ Status StreamLoadAction::_process_put(HttpRequest* http_req, StreamLoadContext*
             return Status::InvalidArgument("Invalid mem limit format");
         }
     }
-    if (!http_req->header(HTTP_EXEC_JSONPATHS).empty()) {
-        request.__set_jsonpaths(http_req->header(HTTP_EXEC_JSONPATHS));
+    if (!http_req->header(HTTP_JSONPATHS).empty()) {
+        request.__set_jsonpaths(http_req->header(HTTP_JSONPATHS));
     }
-    if (!http_req->header(HTTP_EXEC_STRIP_OUTER_ARRAY).empty()) {
-        if (boost::iequals(http_req->header(HTTP_EXEC_STRIP_OUTER_ARRAY), "true")) {
+    if (!http_req->header(HTTP_JSONROOT).empty()) {
+       request.__set_json_root(http_req->header(HTTP_JSONROOT));
+    }
+    if (!http_req->header(HTTP_STRIP_OUTER_ARRAY).empty()) {
+        if (boost::iequals(http_req->header(HTTP_STRIP_OUTER_ARRAY), "true")) {
             request.__set_strip_outer_array(true);
         } else {
             request.__set_strip_outer_array(false);
diff --git a/be/src/http/http_common.h b/be/src/http/http_common.h
index b1001f9eec..984aa102a3 100644
--- a/be/src/http/http_common.h
+++ b/be/src/http/http_common.h
@@ -36,8 +36,9 @@ static const std::string HTTP_NEGATIVE = "negative";
 static const std::string HTTP_STRICT_MODE = "strict_mode";
 static const std::string HTTP_TIMEZONE = "timezone";
 static const std::string HTTP_EXEC_MEM_LIMIT = "exec_mem_limit";
-static const std::string HTTP_EXEC_JSONPATHS  = "jsonpaths";
-static const std::string HTTP_EXEC_STRIP_OUTER_ARRAY = "strip_outer_array";
+static const std::string HTTP_JSONPATHS  = "jsonpaths";
+static const std::string HTTP_JSONROOT  = "json_root";
+static const std::string HTTP_STRIP_OUTER_ARRAY = "strip_outer_array";
 
 static const std::string HTTP_100_CONTINUE = "100-continue";
 
diff --git a/be/test/exec/json_scanner_test.cpp b/be/test/exec/json_scanner_test.cpp
index 1deaca2b26..c49282fc0e 100644
--- a/be/test/exec/json_scanner_test.cpp
+++ b/be/test/exec/json_scanner_test.cpp
@@ -401,6 +401,8 @@ TEST_F(JsonSannerTest, normal_simple_arrayjson) {
         range.start_offset = 0;
         range.size = -1;
         range.format_type = TFileFormatType::FORMAT_JSON;
+        range.strip_outer_array = true;
+        range.__isset.strip_outer_array = true;
         range.splittable = true;
         range.path = "./be/test/exec/test_data/json_scanner/test_simple2.json";
         range.file_type = TFileType::FILE_LOCAL;
diff --git a/docs/en/sql-reference/sql-statements/Data Manipulation/ROUTINE LOAD.md b/docs/en/sql-reference/sql-statements/Data Manipulation/ROUTINE LOAD.md
index 2faf51c7f7..9772f5b4f5 100644
--- a/docs/en/sql-reference/sql-statements/Data Manipulation/ROUTINE LOAD.md	
+++ b/docs/en/sql-reference/sql-statements/Data Manipulation/ROUTINE LOAD.md	
@@ -178,6 +178,9 @@ FROM data_source
     8. `strip_outer_array`
         Boolean type, true to indicate that json data starts with an array object and flattens objects in the array object, default value is false.
 
+    9. `json_root`
+        json_root is a valid JSONPATH string that specifies the root node of the JSON Document. The default value is "".
+
 5. data_source
 
     The type of data source. Current support:
@@ -461,6 +464,36 @@ FROM data_source
      1）If the json data starts as an array and each object in the array is a record, you need to set the strip_outer_array to true to represent the flat array.
      2）If the json data starts with an array, and each object in the array is a record, our ROOT node is actually an object in the array when we set jsonpath.
 
+6. User specifies the json_root node
+        CREATE ROUTINE LOAD example_db.test1 ON example_tbl
+        COLUMNS(category, author, price, timestamp, dt=from_unixtime(timestamp, '%Y%m%d'))
+        PROPERTIES
+        (
+            "desired_concurrent_number"="3",
+            "max_batch_interval" = "20",
+            "max_batch_rows" = "300000",
+            "max_batch_size" = "209715200",
+            "strict_mode" = "false",
+            "format" = "json",
+            "jsonpaths" = "[\"$.category\",\"$.author\",\"$.price\",\"$.timestamp\"]",
+            "strip_outer_array" = "true",
+            "json_root" = "$.RECORDS"
+        )
+        FROM KAFKA
+        (
+            "kafka_broker_list" = "broker1:9092,broker2:9092,broker3:9092",
+            "kafka_topic" = "my_topic",
+            "kafka_partitions" = "0,1,2",
+            "kafka_offsets" = "0,0,0"
+        );
+   For example json data:
+        {
+        "RECORDS":[
+            {"category":"11","title":"SayingsoftheCentury","price":895,"timestamp":1589191587},
+            {"category":"22","author":"2avc","price":895,"timestamp":1589191487},
+            {"category":"33","author":"3avc","title":"SayingsoftheCentury","timestamp":1589191387}
+            ]
+        }
 ## keyword
 
     CREATE, ROUTINE, LOAD
diff --git a/docs/en/sql-reference/sql-statements/Data Manipulation/STREAM LOAD.md b/docs/en/sql-reference/sql-statements/Data Manipulation/STREAM LOAD.md
index 17c88a4e4d..03ed5db392 100644
--- a/docs/en/sql-reference/sql-statements/Data Manipulation/STREAM LOAD.md	
+++ b/docs/en/sql-reference/sql-statements/Data Manipulation/STREAM LOAD.md	
@@ -120,6 +120,9 @@ There are two ways to import json: simple mode and matched mode. If jsonpath is
 `strip_outer_array`
 Boolean type, true to indicate that json data starts with an array object and flattens objects in the array object, default value is false.
 
+`json_root`
+json_root is a valid JSONPATH string that specifies the root node of the JSON Document. The default value is "".
+
 RETURN VALUES
 
 After the load is completed, the related content of this load will be returned in Json format. Current field included
@@ -213,7 +216,7 @@ Where url is the url given by ErrorURL.
                {"category":"Linux","author":"avc","title":"Linux kernel","price":195}
               ]
 11. Matched load json by jsonpaths
-       json data：
+       For example json data:
            [
            {"category":"xuxb111","author":"1avc","title":"SayingsoftheCentury","price":895},
            {"category":"xuxb222","author":"2avc","title":"SayingsoftheCentury","price":895},
@@ -225,6 +228,18 @@ Where url is the url given by ErrorURL.
         1）If the json data starts as an array and each object in the array is a record, you need to set the strip_outer_array to true to represent the flat array.
         2）If the json data starts with an array, and each object in the array is a record, our ROOT node is actually an object in the array when we set jsonpath.
 
+12. User specifies the json_root node
+       For example json data:
+            {
+            "RECORDS":[
+                {"category":"11","title":"SayingsoftheCentury","price":895,"timestamp":1589191587},
+                {"category":"22","author":"2avc","price":895,"timestamp":1589191487},
+                {"category":"33","author":"3avc","title":"SayingsoftheCentury","timestamp":1589191387}
+                ]
+            }
+       Matched imports are made by specifying jsonpath parameter, such as `category`, `author`, and `price`, for example: 
+         curl --location-trusted -u root  -H "columns: category, price, author" -H "label:123" -H "format: json" -H "jsonpaths: [\"$.category\",\"$.price\",\"$.author\"]" -H "strip_outer_array: true" -H "json_root: $.RECORDS" -T testData http://host:port/api/testDb/testTbl/_stream_load
+
 ## keyword
 
     STREAM, LOAD
diff --git a/docs/zh-CN/sql-reference/sql-statements/Data Manipulation/ROUTINE LOAD.md b/docs/zh-CN/sql-reference/sql-statements/Data Manipulation/ROUTINE LOAD.md
index 3a870e57b6..a84ab43ffc 100644
--- a/docs/zh-CN/sql-reference/sql-statements/Data Manipulation/ROUTINE LOAD.md	
+++ b/docs/zh-CN/sql-reference/sql-statements/Data Manipulation/ROUTINE LOAD.md	
@@ -158,6 +158,10 @@ under the License.
 
             布尔类型，为true表示json数据以数组对象开始且将数组对象中进行展平，默认值是false。
 
+        9. json_root
+
+        json_root为合法的jsonpath字符串，用于指定json document的根节点，默认值为""。
+
     5. data_source
 
         数据源的类型。当前支持：
@@ -398,6 +402,36 @@ under the License.
            1）如果json数据是以数组开始，并且数组中每个对象是一条记录，则需要将strip_outer_array设置成true，表示展平数组。
            2）如果json数据是以数组开始，并且数组中每个对象是一条记录，在设置jsonpath时，我们的ROOT节点实际上是数组中对象。
 
+    6. 用户指定根节点json_root
+        CREATE ROUTINE LOAD example_db.test1 ON example_tbl
+        COLUMNS(category, author, price, timestamp, dt=from_unixtime(timestamp, '%Y%m%d'))
+        PROPERTIES
+        (
+            "desired_concurrent_number"="3",
+            "max_batch_interval" = "20",
+            "max_batch_rows" = "300000",
+            "max_batch_size" = "209715200",
+            "strict_mode" = "false",
+            "format" = "json",
+            "jsonpaths" = "[\"$.category\",\"$.author\",\"$.price\",\"$.timestamp\"]",
+            "strip_outer_array" = "true",
+            "json_root" = "$.RECORDS"
+        )
+        FROM KAFKA
+        (
+            "kafka_broker_list" = "broker1:9092,broker2:9092,broker3:9092",
+            "kafka_topic" = "my_topic",
+            "kafka_partitions" = "0,1,2",
+            "kafka_offsets" = "0,0,0"
+        );
+   json数据格式:
+        {
+        "RECORDS":[
+            {"category":"11","title":"SayingsoftheCentury","price":895,"timestamp":1589191587},
+            {"category":"22","author":"2avc","price":895,"timestamp":1589191487},
+            {"category":"33","author":"3avc","title":"SayingsoftheCentury","timestamp":1589191387}
+            ]
+        }
 ## keyword
 
     CREATE,ROUTINE,LOAD
diff --git a/docs/zh-CN/sql-reference/sql-statements/Data Manipulation/STREAM LOAD.md b/docs/zh-CN/sql-reference/sql-statements/Data Manipulation/STREAM LOAD.md
index 68f895f5e6..b95912d288 100644
--- a/docs/zh-CN/sql-reference/sql-statements/Data Manipulation/STREAM LOAD.md	
+++ b/docs/zh-CN/sql-reference/sql-statements/Data Manipulation/STREAM LOAD.md	
@@ -89,6 +89,8 @@ under the License.
            ]
            当strip_outer_array为true，最后导入到doris中会生成两行数据。
 
+        json_root: json_root为合法的jsonpath字符串，用于指定json document的根节点，默认值为""。
+
     RETURN VALUES
         导入完成后，会以Json格式返回这次导入的相关内容。当前包括一下字段
         Status: 导入最后的状态。
@@ -172,6 +174,18 @@ under the License.
            1）如果json数据是以数组开始，并且数组中每个对象是一条记录，则需要将strip_outer_array设置成true，表示展平数组。
            2）如果json数据是以数组开始，并且数组中每个对象是一条记录，在设置jsonpath时，我们的ROOT节点实际上是数组中对象。
 
+    12. 用户指定json根节点
+       json数据格式:
+            {
+            "RECORDS":[
+                {"category":"11","title":"SayingsoftheCentury","price":895,"timestamp":1589191587},
+                {"category":"22","author":"2avc","price":895,"timestamp":1589191487},
+                {"category":"33","author":"3avc","title":"SayingsoftheCentury","timestamp":1589191387}
+                ]
+            }
+        通过指定jsonpath进行精准导入，例如只导入category、author、price三个属性  
+         curl --location-trusted -u root  -H "columns: category, price, author" -H "label:123" -H "format: json" -H "jsonpaths: [\"$.category\",\"$.price\",\"$.author\"]" -H "strip_outer_array: true" -H "json_root: $.RECORDS" -T testData http://host:port/api/testDb/testTbl/_stream_load
+
 ## keyword
     STREAM,LOAD
 
diff --git a/fe/fe-core/src/main/java/org/apache/doris/analysis/CreateRoutineLoadStmt.java b/fe/fe-core/src/main/java/org/apache/doris/analysis/CreateRoutineLoadStmt.java
index 6fe4d87da7..6861cafdc1 100644
--- a/fe/fe-core/src/main/java/org/apache/doris/analysis/CreateRoutineLoadStmt.java
+++ b/fe/fe-core/src/main/java/org/apache/doris/analysis/CreateRoutineLoadStmt.java
@@ -92,6 +92,7 @@ public class CreateRoutineLoadStmt extends DdlStmt {
     public static final String FORMAT = "format";// the value is csv or json, default is csv
     public static final String STRIP_OUTER_ARRAY = "strip_outer_array";
     public static final String JSONPATHS = "jsonpaths";
+    public static final String JSONROOT = "json_root";
 
     // kafka type properties
     public static final String KAFKA_BROKER_LIST_PROPERTY = "kafka_broker_list";
@@ -113,6 +114,7 @@ public class CreateRoutineLoadStmt extends DdlStmt {
             .add(FORMAT)
             .add(JSONPATHS)
             .add(STRIP_OUTER_ARRAY)
+            .add(JSONROOT)
             .add(LoadStmt.STRICT_MODE)
             .add(LoadStmt.TIMEZONE)
             .build();
@@ -149,11 +151,11 @@ public class CreateRoutineLoadStmt extends DdlStmt {
      *   1) dataFormat = "json"
      *   2) jsonPaths = "$.XXX.xxx"
      */
-    private String format   = ""; //default is csv.
-    private String jsonPaths     = "";
+    private String format     = ""; //default is csv.
+    private String jsonPaths  = "";
+    private String jsonRoot   = ""; // MUST be a jsonpath string
     private boolean stripOuterArray = false;
 
-
     // kafka related properties
     private String kafkaBrokerList;
     private String kafkaTopic;
@@ -240,6 +242,10 @@ public class CreateRoutineLoadStmt extends DdlStmt {
         return jsonPaths;
     }
 
+    public String getJsonRoot() {
+        return jsonRoot;
+    }
+
     public String getKafkaBrokerList() {
         return kafkaBrokerList;
     }
@@ -364,6 +370,7 @@ public class CreateRoutineLoadStmt extends DdlStmt {
             } else if (format.equalsIgnoreCase("json")) {
                 format = "json";
                 jsonPaths = jobProperties.get(JSONPATHS);
+                jsonRoot = jobProperties.get(JSONROOT);
                 stripOuterArray = Boolean.valueOf(jobProperties.get(STRIP_OUTER_ARRAY));
             } else {
                 throw new UserException("Format type is invalid. format=`" + format + "`");
diff --git a/fe/fe-core/src/main/java/org/apache/doris/load/routineload/RoutineLoadJob.java b/fe/fe-core/src/main/java/org/apache/doris/load/routineload/RoutineLoadJob.java
index 5e236c8808..feaf50f171 100644
--- a/fe/fe-core/src/main/java/org/apache/doris/load/routineload/RoutineLoadJob.java
+++ b/fe/fe-core/src/main/java/org/apache/doris/load/routineload/RoutineLoadJob.java
@@ -61,7 +61,6 @@ import org.apache.doris.transaction.AbstractTxnStateChangeCallback;
 import org.apache.doris.transaction.TransactionException;
 import org.apache.doris.transaction.TransactionState;
 import org.apache.doris.transaction.TransactionStatus;
-
 import com.google.common.base.Joiner;
 import com.google.common.base.Preconditions;
 import com.google.common.base.Strings;
@@ -70,7 +69,6 @@ import com.google.common.collect.Lists;
 import com.google.common.collect.Maps;
 import com.google.gson.Gson;
 import com.google.gson.GsonBuilder;
-
 import org.apache.logging.log4j.LogManager;
 import org.apache.logging.log4j.Logger;
 
@@ -184,6 +182,7 @@ public abstract class RoutineLoadJob extends AbstractTxnStateChangeCallback impl
     private static final String PROPS_FORMAT = "format";
     private static final String PROPS_STRIP_OUTER_ARRAY = "strip_outer_array";
     private static final String PROPS_JSONPATHS = "jsonpaths";
+    private static final String PROPS_JSONROOT = "json_root";
 
     protected int currentTaskConcurrentNum;
     protected RoutineLoadProgress progress;
@@ -285,6 +284,7 @@ public abstract class RoutineLoadJob extends AbstractTxnStateChangeCallback impl
             jobProperties.put(PROPS_FORMAT, "csv");
             jobProperties.put(PROPS_STRIP_OUTER_ARRAY, "false");
             jobProperties.put(PROPS_JSONPATHS, "");
+            jobProperties.put(PROPS_JSONROOT, "");
         } else if (stmt.getFormat().equals("json")) {
             jobProperties.put(PROPS_FORMAT, "json");
             if (!Strings.isNullOrEmpty(stmt.getJsonPaths())) {
@@ -292,6 +292,11 @@ public abstract class RoutineLoadJob extends AbstractTxnStateChangeCallback impl
             } else {
                 jobProperties.put(PROPS_JSONPATHS, "");
             }
+            if (!Strings.isNullOrEmpty(stmt.getJsonRoot())) {
+                jobProperties.put(PROPS_JSONROOT, stmt.getJsonRoot());
+            } else {
+                jobProperties.put(PROPS_JSONROOT, "");
+            }
             if (stmt.isStripOuterArray()) {
                 jobProperties.put(PROPS_STRIP_OUTER_ARRAY, "true");
             } else {
@@ -472,6 +477,14 @@ public abstract class RoutineLoadJob extends AbstractTxnStateChangeCallback impl
         return value;
     }
 
+    public String getJsonRoot() {
+        String value = jobProperties.get(PROPS_JSONROOT);
+        if (value == null) {
+            return "";
+        }
+        return value;
+    }
+
     public int getSizeOfRoutineLoadTaskInfoList() {
         readLock();
         try {
diff --git a/fe/fe-core/src/main/java/org/apache/doris/planner/StreamLoadScanNode.java b/fe/fe-core/src/main/java/org/apache/doris/planner/StreamLoadScanNode.java
index 2fa49190b0..5725af2a7b 100644
--- a/fe/fe-core/src/main/java/org/apache/doris/planner/StreamLoadScanNode.java
+++ b/fe/fe-core/src/main/java/org/apache/doris/planner/StreamLoadScanNode.java
@@ -103,6 +103,9 @@ public class StreamLoadScanNode extends LoadScanNode {
             if (!streamLoadTask.getJsonPaths().isEmpty()) {
                 rangeDesc.setJsonpaths(streamLoadTask.getJsonPaths());
             }
+            if (!streamLoadTask.getJsonRoot().isEmpty()) {
+                rangeDesc.setJson_root(streamLoadTask.getJsonRoot());
+            }
             rangeDesc.setStrip_outer_array(streamLoadTask.isStripOuterArray());
         }
         rangeDesc.splittable = false;
diff --git a/fe/fe-core/src/main/java/org/apache/doris/task/StreamLoadTask.java b/fe/fe-core/src/main/java/org/apache/doris/task/StreamLoadTask.java
index b09b3478fe..d5ff24e137 100644
--- a/fe/fe-core/src/main/java/org/apache/doris/task/StreamLoadTask.java
+++ b/fe/fe-core/src/main/java/org/apache/doris/task/StreamLoadTask.java
@@ -53,6 +53,7 @@ public class StreamLoadTask {
     private TFileFormatType formatType;
     private boolean stripOuterArray;
     private String jsonPaths;
+    private String jsonRoot;
 
     // optional
     private List<ImportColumnDesc> columnExprDescs = Lists.newArrayList();
@@ -72,6 +73,7 @@ public class StreamLoadTask {
         this.fileType = fileType;
         this.formatType = formatType;
         this.jsonPaths = "";
+        this.jsonRoot = "";
         this.stripOuterArray = false;
     }
 
@@ -143,6 +145,14 @@ public class StreamLoadTask {
         this.jsonPaths = jsonPaths;
     }
 
+    public String getJsonRoot() {
+        return jsonRoot;
+    }
+
+    public void setJsonRoot(String jsonRoot) {
+        this.jsonRoot = jsonRoot;
+    }
+
     public static StreamLoadTask fromTStreamLoadPutRequest(TStreamLoadPutRequest request, Database db) throws UserException {
         StreamLoadTask streamLoadTask = new StreamLoadTask(request.getLoadId(), request.getTxnId(),
                                                            request.getFileType(), request.getFormatType());
@@ -194,6 +204,9 @@ public class StreamLoadTask {
             if (request.getJsonpaths() != null) {
                 jsonPaths = request.getJsonpaths();
             }
+            if (request.getJson_root() != null) {
+                jsonRoot = request.getJson_root();
+            }
             stripOuterArray = request.isStrip_outer_array();
         }
     }
@@ -225,6 +238,9 @@ public class StreamLoadTask {
         if (!routineLoadJob.getJsonPaths().isEmpty()) {
             jsonPaths = routineLoadJob.getJsonPaths();
         }
+        if (!routineLoadJob.getJsonRoot().isEmpty()) {
+            jsonRoot = routineLoadJob.getJsonRoot();
+        }
         stripOuterArray = routineLoadJob.isStripOuterArray();
     }
 
diff --git a/gensrc/thrift/FrontendService.thrift b/gensrc/thrift/FrontendService.thrift
index 28ccb28cf2..dd5b6ec892 100644
--- a/gensrc/thrift/FrontendService.thrift
+++ b/gensrc/thrift/FrontendService.thrift
@@ -551,6 +551,7 @@ struct TStreamLoadPutRequest {
     23: optional bool strip_outer_array
     24: optional string jsonpaths
     25: optional i64 thrift_rpc_timeout_ms
+    26: optional string json_root
 }
 
 struct TStreamLoadPutResult {
diff --git a/gensrc/thrift/PlanNodes.thrift b/gensrc/thrift/PlanNodes.thrift
index eb64042c8b..3ebff28f1e 100644
--- a/gensrc/thrift/PlanNodes.thrift
+++ b/gensrc/thrift/PlanNodes.thrift
@@ -129,6 +129,7 @@ struct TBrokerRangeDesc {
     //  it's usefull when format_type == FORMAT_JSON
     11: optional bool strip_outer_array;
     12: optional string jsonpaths;
+    13: optional string json_root;
 }
 
 struct TBrokerScanRangeParams {