diff --git a/.gitignore b/.gitignore
index d1f1dc4dedb31da5b41a1046d7a4419881746071..b0a36fdeddeb2815d5ff5ba1c6a2241f13f678f5 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,9 +1,12 @@
 build/
 .ycm_extra_conf.py
 .vscode/
+.cache/
+compile_commands.json
 .idea/
 cmake-build-debug/
 cmake-build-release/
+cmake-build-relwithdebinfo/
 cscope.out
 cscope.files
 tags
diff --git a/packaging/cfg/taos.cfg b/packaging/cfg/taos.cfg
index a6cbe6f3fb5278c230b40ab0d814932888a7e548..9b07ed7ab3c2ead684d8318c2a13989bc87ad7b9 100644
--- a/packaging/cfg/taos.cfg
+++ b/packaging/cfg/taos.cfg
@@ -313,3 +313,12 @@ keepColumnName            1
 
 # unit Hour. Latency of data migration
 # keepTimeOffset     0
+
+# taosc write batch size, maximum 4096, suggested value 64 ~ 512, default 0, 0 means disable write batching.
+# writeBatchSize        0
+
+# taosc write batch timeout in milliseconds, maximum 2048, suggested value 2 ~ 100, default 10.
+# writeBatchTimeout     10
+
+# using thread local write batching. this option is not available when writeBatchSize = 0.
+# writeBatchThreadLocal 0
\ No newline at end of file
diff --git a/src/client/inc/tscBatchMerge.h b/src/client/inc/tscBatchMerge.h
new file mode 100644
index 0000000000000000000000000000000000000000..12fe66c9cf6df4ee1d0be24def41d7e05efd8b82
--- /dev/null
+++ b/src/client/inc/tscBatchMerge.h
@@ -0,0 +1,274 @@
+/*
+* Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+*
+* This program is free software: you can use, redistribute, and/or modify
+* it under the terms of the GNU Affero General Public License, version 3
+* or later ("AGPL"), as published by the Free Software Foundation.
+*
+* This program is distributed in the hope that it will be useful, but WITHOUT
+* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+* FITNESS FOR A PARTICULAR PURPOSE.
+*
+* You should have received a copy of the GNU Affero General Public License
+* along with this program. If not, see <http://www.gnu.org/licenses/>.
+*/
+
+#ifndef TDENGINE_TSCBATCHMERGE_H
+#define TDENGINE_TSCBATCHMERGE_H
+
+#include "hash.h"
+#include "taosmsg.h"
+#include "tarray.h"
+#include "tscUtil.h"
+#include "tsclient.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/**
+ * A builder of SSubmitBlk.
+ */
+typedef struct SSubmitBlkBuilder {
+  // the metadata of the SSubmitBlk.
+  SSubmitBlk* metadata;
+  
+  // the array stores all the rows in a table, aka SArray<SMemRow>.
+  SArray* rows;
+  
+} SSubmitBlkBuilder;
+
+/**
+ * The builder to build SSubmitMsg::blocks.
+ */
+typedef struct SSubmitMsgBlocksBuilder {
+  // SHashObj<table_uid, SSubmitBlkBuilder*>.
+  SHashObj* blockBuilders;
+  int64_t vgId;
+} SSubmitMsgBlocksBuilder;
+
+/**
+ * STableDataBlocksBuilder is a tool to build data blocks by append the existing data blocks in a vnode.
+ */
+typedef struct STableDataBlocksBuilder {
+  SSubmitMsgBlocksBuilder* blocksBuilder;
+  STableDataBlocks* firstBlock;
+  int64_t vgId;
+} STableDataBlocksBuilder;
+
+/**
+ * STableDataBlocksListBuilder is a tool to build vnode data blocks list by appending exist data blocks.
+ */
+typedef struct STableDataBlocksListBuilder {
+  SHashObj* dataBlocksBuilders;
+} STableDataBlocksListBuilder;
+
+/**
+ * A Builder to build SInsertStatementParam::pTableNameList.
+ */
+typedef struct STableNameListBuilder {
+  // store the unsorted table names, SArray<SName*>.
+  SArray* pTableNameList;
+} STableNameListBuilder;
+
+/**
+ * Create a SSubmitBlkBuilder using exist metadata.
+ * 
+ * @param metadata  the metadata.
+ * @return          the SSubmitBlkBuilder.
+ */
+SSubmitBlkBuilder* createSSubmitBlkBuilder(SSubmitBlk* metadata);
+
+/**
+ * Destroy the SSubmitBlkBuilder.
+ * 
+ * @param builder the SSubmitBlkBuilder.
+ */
+void destroySSubmitBlkBuilder(SSubmitBlkBuilder* builder);
+
+/**
+ * Append a SSubmitBlk* to the builder. The table uid in pBlock must be the same with the builder's.
+ * 
+ * @param builder  the SSubmitBlkBuilder.
+ * @param pBlock   the pBlock to append.
+ * @return         whether the append is success.
+ */
+bool appendSSubmitBlkBuilder(SSubmitBlkBuilder* builder, SSubmitBlk *pBlock);
+
+
+/**
+ * Build and write SSubmitBlk to `target`
+ * 
+ * @param builder   the SSubmitBlkBuilder.
+ * @param target    the target to write.
+ * @param nRows     the number of rows in SSubmitBlk*.
+ * @return          the writen bytes.
+ */
+size_t writeSSubmitBlkBuilder(SSubmitBlkBuilder* builder, SSubmitBlk* target, size_t* nRows);
+
+/**
+ * Get the expected writen bytes of `writeSSubmitBlkBuilder`.
+ * 
+ * @param builder  the SSubmitBlkBuilder.
+ * @return         the expected writen bytes of `writeSSubmitBlkBuilder`.
+ */
+size_t nWriteSSubmitBlkBuilder(SSubmitBlkBuilder* builder);
+
+/**
+ * Create a SSubmitMsgBuilder.
+ * 
+ * @param vgId  the vgId of SSubmitMsg.
+ * @return      the SSubmitMsgBuilder.
+ */
+SSubmitMsgBlocksBuilder* createSSubmitMsgBuilder(int64_t vgId);
+
+/**
+ * Get the expected writen bytes of `writeSSubmitMsgBlocksBuilder`.
+ * 
+ * @param builder  the SSubmitMsgBlocksBuilder.
+ * @return         the expected writen bytes of `writeSSubmitMsgBlocksBuilder`.
+ */
+size_t nWriteSSubmitMsgBuilder(SSubmitMsgBlocksBuilder* builder);
+
+/**
+ * Build and write SSubmitMsg::blocks to `pBlocks`
+ * 
+ * @param builder   the SSubmitBlkBuilder.
+ * @param pBlocks   the target to write.
+ * @param nRows     the number of row in SSubmitMsg::blocks.
+ * @return          the writen bytes.
+ */
+size_t writeSSubmitMsgBlocksBuilder(SSubmitMsgBlocksBuilder* builder, SSubmitBlk* pBlocks, size_t* nRows);
+
+/**
+ * Get the number of block in SSubmitMsgBlocksBuilder.
+ * @param builder the SSubmitMsgBlocksBuilder.
+ * @return        the number of SSubmitBlk block.
+ */
+size_t nBlockSSubmitMsgBlocksBuilder(SSubmitMsgBlocksBuilder* builder);
+
+/**
+ * Destroy the SSubmitMsgBlocksBuilder.
+ * 
+ * @param builder the SSubmitMsgBlocksBuilder to destroy.
+ */
+void destroySSubmitMsgBuilder(SSubmitMsgBlocksBuilder* builder);
+
+/**
+ * Append SSubmitMsg* to the SSubmitMsgBlocksBuilder.
+ * 
+ * @param builder       the SSubmitMsgBlocksBuilder.
+ * @param pBlocks       the SSubmitBlk in SSubmitMsg::blocks. 
+ * @param nBlocks       the number of blocks in SSubmitMsg.
+ * @return whether the append is success.
+ */
+bool appendSSubmitMsgBlocks(SSubmitMsgBlocksBuilder* builder, SSubmitBlk* pBlocks, size_t nBlocks);
+
+/**
+ * Create the STableDataBlocksBuilder.
+ * 
+ * @param vgId the vgId of STableDataBlocksBuilder.
+ * @return the STableDataBlocksBuilder.
+ */
+STableDataBlocksBuilder* createSTableDataBlocksBuilder(int64_t vgId);
+
+/**
+ * Destroy the STableDataBlocksBuilder.
+ * @param builder the STableDataBlocksBuilder.
+ */
+void destroySTableDataBlocksBuilder(STableDataBlocksBuilder *builder);
+
+/**
+ * Append a data blocks to STableDataBlocksBuilder.
+ * @param builder    the STableDataBlocksBuilder.
+ * @param dataBlocks the dataBlocks to append. the vgId of dataBlocks must be same with the STableDataBlocksBuilder.
+ * @return           whether the append is success.
+ */
+bool appendSTableDataBlocksBuilder(STableDataBlocksBuilder* builder, STableDataBlocks* dataBlocks);
+
+/**
+ * Build the data blocks for single vnode.
+ * @param builder   the STableDataBlocksBuilder.
+ * @param nRows     the number of row in STableDataBlocks.
+ * @return          the data blocks for single vnode.
+ */
+STableDataBlocks* buildSTableDataBlocksBuilder(STableDataBlocksBuilder* builder, size_t* nRows);
+
+/**
+ * Create the STableDataBlocksListBuilder.
+ * 
+ * @return the STableDataBlocksListBuilder.
+ */
+STableDataBlocksListBuilder* createSTableDataBlocksListBuilder();
+
+/**
+ * Destroy the STableDataBlocksListBuilder.
+ * 
+ * @param builder the STableDataBlocksListBuilder.
+ */
+void destroySTableDataBlocksListBuilder(STableDataBlocksListBuilder* builder);
+
+/**
+ * Append a data blocks to STableDataBlocksListBuilder.
+ * 
+ * @param builder       the STableDataBlocksListBuilder.
+ * @param dataBlocks    the data blocks.
+ * @return              whether the append is success.
+ */
+bool appendSTableDataBlocksListBuilder(STableDataBlocksListBuilder* builder, STableDataBlocks* dataBlocks);
+
+/**
+ * Build the vnode data blocks list.
+ * 
+ * @param builder   the STableDataBlocksListBuilder.
+ * @param nTables   the number of table in vnode data blocks list.
+ * @param nRows     the number of row in vnode data blocks list. 
+ * @return          the vnode data blocks list.
+ */
+SArray* buildSTableDataBlocksListBuilder(STableDataBlocksListBuilder* builder, size_t* nTables, size_t* nRows);
+
+/**
+ * Create STableNameListBuilder.
+ */
+STableNameListBuilder* createSTableNameListBuilder();
+
+/**
+ * Destroy the STableNameListBuilder.
+ * @param builder the STableNameListBuilder.
+ */
+void destroySTableNameListBuilder(STableNameListBuilder* builder);
+
+/**
+ * Insert a SName to builder.
+ * 
+ * @param builder  the STableNameListBuilder.
+ * @param name     the table name.
+ * @return         whether it is success.
+ */
+bool insertSTableNameListBuilder(STableNameListBuilder* builder, SName* name);
+
+/**
+ * Build the STable name list.
+ * 
+ * @param builder       the STableNameListBuilder.
+ * @param numOfTables   the number of table.
+ * @return              the STable name list.
+ */
+SName** buildSTableNameListBuilder(STableNameListBuilder* builder, size_t* numOfTables);
+
+/**
+ * Merge the KV-PayLoad SQL objects into single one. 
+ * The statements here must be an insertion statement and no schema attached.
+ * 
+ * @param polls  the array of SSqlObj*.
+ * @param nPolls the number of SSqlObj* in the array.
+ * @param result the returned result. result is not null!
+ * @return       the status code.
+ */
+int32_t tscMergeSSqlObjs(SSqlObj** polls, size_t nPolls, SSqlObj *result);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif  // TDENGINE_TSCBATCHMERGE_H
diff --git a/src/client/inc/tscBatchWrite.h b/src/client/inc/tscBatchWrite.h
new file mode 100644
index 0000000000000000000000000000000000000000..0e7d0136b068c30ced3cbe158bc0e60bc660cae4
--- /dev/null
+++ b/src/client/inc/tscBatchWrite.h
@@ -0,0 +1,235 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef TDENGINE_TSCBATCHWRITE_H
+#define TDENGINE_TSCBATCHWRITE_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "tthread.h"
+
+// forward declaration.
+typedef struct STscObj                   STscObj;
+typedef struct SSqlObj                   SSqlObj;
+typedef struct SDispatcherTimeoutManager SDispatcherTimeoutManager;
+
+/**
+ * SAsyncBatchWriteDispatcher is an async batching write dispatcher (ABWD). ABWD accepts the recent SQL requests and put
+ * them in a queue waiting to be scheduled. When the number of requests in the queue reaches batch_size, it merges the
+ * requests in the queue and sends them to the server, thus reducing the network overhead caused by multiple
+ * communications to the server and directly improving the throughput of small object asynchronous writes.
+ */
+typedef struct SAsyncBatchWriteDispatcher {
+  // the client object.
+  STscObj* pClient;
+
+  // the timeout manager.
+  SDispatcherTimeoutManager* timeoutManager;
+
+  // the mutex to protect the dispatcher.
+  pthread_mutex_t bufferMutex;
+
+  // the cond to signal when buffer not full.
+  pthread_cond_t notFull;
+
+  // the maximum number of insertion rows in a batch.
+  int32_t batchSize;
+
+  // the number of insertion rows in the buffer.
+  int32_t currentSize;
+
+  // the number of items in the buffer.
+  int32_t bufferSize;
+
+  // whether the dispatcher is shutdown.
+  bool shutdown;
+
+  SSqlObj* buffer[];
+} SAsyncBatchWriteDispatcher;
+
+/**
+ * The manager of SAsyncBatchWriteDispatcher. Call dispatcherAcquire(...) to get the SAsyncBatchWriteDispatcher
+ * instance. SDispatcherManager will manage the life cycle of SAsyncBatchWriteDispatcher.
+ */
+typedef struct SDispatcherManager {
+  pthread_key_t key;
+
+  // the maximum number of insertion rows in a batch.
+  int32_t batchSize;
+
+  // the batching timeout in milliseconds.
+  int32_t timeoutMs;
+
+  // specifies whether the dispatcher is thread local, if the dispatcher is not
+  // thread local, we will use the global dispatcher below.
+  bool isThreadLocal;
+
+  // the global dispatcher, if thread local enabled, global will be set to NULL.
+  SAsyncBatchWriteDispatcher* pGlobal;
+
+  // the client object.
+  STscObj* pClient;
+
+} SDispatcherManager;
+
+/**
+ * Control the timeout of the dispatcher queue.
+ */
+typedef struct SDispatcherTimeoutManager {
+  // the dispatcher that timeout manager belongs to.
+  SAsyncBatchWriteDispatcher* dispatcher;
+
+  // the background thread.
+  pthread_t background;
+
+  // the mutex to sleep the background thread.
+  pthread_mutex_t sleepMutex;
+
+  // the cond to signal to background thread.
+  pthread_cond_t timeout;
+
+  // the batching timeout in milliseconds.
+  int32_t timeoutMs;
+
+  // whether the timeout manager is shutdown.
+  bool shutdown;
+} SDispatcherTimeoutManager;
+
+/**
+ * A batch that polls from SAsyncBatchWriteDispatcher::buffer.
+ */
+typedef struct SBatchRequest {
+  size_t   nRequests;
+  SSqlObj* pRequests[];
+} SBatchRequest;
+
+/**
+ * Create the dispatcher timeout manager.
+ */
+SDispatcherTimeoutManager* createSDispatcherTimeoutManager(SAsyncBatchWriteDispatcher* dispatcher, int32_t timeoutMs);
+
+/**
+ * Destroy the dispatcher timeout manager.
+ */
+void destroySDispatcherTimeoutManager(SDispatcherTimeoutManager* manager);
+
+/**
+ * Check if the timeout manager is shutdown.
+ * @param manager   the timeout manager.
+ * @return          whether the timeout manager is shutdown.
+ */
+bool isShutdownSDispatcherTimeoutManager(SDispatcherTimeoutManager* manager);
+
+/**
+ * Shutdown the SDispatcherTimeoutManager.
+ * @param manager the SDispatcherTimeoutManager.
+ */
+void shutdownSDispatcherTimeoutManager(SDispatcherTimeoutManager* manager);
+
+/**
+ * Merge SSqlObjs into single SSqlObj.
+ *
+ * @param pRequest  the batch request.
+ * @param batch     the batch SSqlObj*.
+ * @return          the status code.
+ */
+int32_t dispatcherBatchBuilder(SBatchRequest* pRequest, SSqlObj** batch);
+
+/**
+ * Merge the sql statements and execute the merged sql statement asynchronously.
+ *
+ * @param pRequest the batch request. the request will be promised to free after calling this function.
+ */
+void dispatcherAsyncExecute(SBatchRequest* pRequest);
+
+/**
+ * Merge the sql statements and execute the merged sql statement.
+ *
+ * @param pRequest the batch request. you must call free(pRequest) after calling this function.
+ */
+void dispatcherExecute(SBatchRequest* pRequest);
+
+/**
+ * Create the async batch write dispatcher.
+ *
+ * @param pClient   the client object.
+ * @param batchSize When user submit an insert sql to `taos_query_a`, the SSqlObj* will be buffered instead of executing
+ * it. If the number of the buffered rows reach `batchSize`, all the SSqlObj* will be merged and sent to vnodes.
+ * @param timeout   The SSqlObj* will be sent to vnodes no more than `timeout` milliseconds. But the actual time
+ *                  vnodes received the SSqlObj* depends on the network quality.
+ */
+SAsyncBatchWriteDispatcher* createSAsyncBatchWriteDispatcher(STscObj* pClient, int32_t batchSize, int32_t timeoutMs);
+
+/**
+ * Destroy the async auto batch dispatcher.
+ */
+void destroySAsyncBatchWriteDispatcher(SAsyncBatchWriteDispatcher* dispatcher);
+
+/**
+ * Check if the current sql object can be dispatch by ABWD.
+ * 1. auto batch feature on the sql object must be enabled.
+ * 2. must be an `insert into ... value ...` statement.
+ * 3. the payload type must be kv payload.
+ * 4. no schema attached.
+ *
+ * @param dispatcher    the dispatcher.
+ * @param pSql          the sql object to check.
+ * @return              returns true if the sql object can be dispatch by ABWD.
+ */
+bool dispatcherCanDispatch(SAsyncBatchWriteDispatcher* dispatcher, SSqlObj* pSql);
+
+/**
+ * Try to offer the SSqlObj* to the dispatcher. If the number of row reach `batchSize`, the function
+ * will merge the SSqlObj* in the buffer and send them to the vnodes.
+ *
+ * @param pSql the insert statement to offer.
+ * @return     if offer success, returns true.
+ */
+bool dispatcherTryDispatch(SAsyncBatchWriteDispatcher* dispatcher, SSqlObj* pSql);
+
+/**
+ * Create the manager of SAsyncBatchWriteDispatcher.
+ *
+ * @param pClient       the client object.
+ * @param batchSize     the batchSize of SAsyncBatchWriteDispatcher.
+ * @param timeoutMs     the timeoutMs of SAsyncBatchWriteDispatcher.
+ * @param isThreadLocal specifies whether the dispatcher is thread local.
+ * @return the SAsyncBatchWriteDispatcher manager.
+ */
+SDispatcherManager* createDispatcherManager(STscObj* pClient, int32_t batchSize, int32_t timeoutMs, bool isThreadLocal);
+
+/**
+ * Destroy the SDispatcherManager.
+ * (will destroy all the instances of SAsyncBatchWriteDispatcher in the thread local variable)
+ *
+ * @param manager  the SDispatcherManager.
+ */
+void destroyDispatcherManager(SDispatcherManager* manager);
+
+/**
+ * Get an instance of SAsyncBatchWriteDispatcher.
+ *
+ * @param manager   the SDispatcherManager.
+ * @return          the SAsyncBatchWriteDispatcher instance.
+ */
+SAsyncBatchWriteDispatcher* dispatcherAcquire(SDispatcherManager* manager);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif  // TDENGINE_TSCBATCHWRITE_H
diff --git a/src/client/inc/tscUtil.h b/src/client/inc/tscUtil.h
index 2699f4dbda4f74b204ba89f108b9c2a91d6cb4e9..2e6c99f68cae6d801a55395170df4e25c137505f 100644
--- a/src/client/inc/tscUtil.h
+++ b/src/client/inc/tscUtil.h
@@ -144,6 +144,8 @@ void doRetrieveSubqueryData(SSchedMsg *pMsg);
 SParamInfo* tscAddParamToDataBlock(STableDataBlocks* pDataBlock, char type, uint8_t timePrec, int16_t bytes,
                                    uint32_t offset);
 
+void destroySTableDataBlocksList(SArray* pDataBlocks);
+void destroySTableDataBlocks(STableDataBlocks* pDataBlocks);
 void*   tscDestroyBlockArrayList(SSqlObj* pSql, SArray* pDataBlockList);
 void*  tscDestroyUdfArrayList(SArray* pUdfList);
 void*   tscDestroyBlockHashTable(SSqlObj* pSql, SHashObj* pBlockHashTable, bool removeMeta);
diff --git a/src/client/inc/tsclient.h b/src/client/inc/tsclient.h
index 65854905bdd12ce7452eddada0b051ff7d1cdcbe..9c2a7220c6340c360440c62bdca5f015dc73232a 100644
--- a/src/client/inc/tsclient.h
+++ b/src/client/inc/tsclient.h
@@ -45,6 +45,7 @@ typedef enum {
 
 // forward declaration
 struct SSqlInfo;
+typedef struct SDispatcherManager SDispatcherManager;
 
 typedef void (*__async_cb_func_t)(void *param, TAOS_RES *tres, int32_t numOfRows);
 typedef void (*_freeSqlSupporter)(void **);
@@ -256,7 +257,7 @@ typedef struct SInsertStatementParam {
 
   int32_t      batchSize;               // for parameter ('?') binding and batch processing
   int32_t      numOfParams;
-
+  int32_t      numOfRows;
   int32_t      numOfFiles;
 
   char         msg[512];                // error message
@@ -352,7 +353,8 @@ typedef struct STscObj {
   SRpcCorEpSet      *tscCorMgmtEpSet;
   pthread_mutex_t    mutex;
   int32_t            numOfObj; // number of sqlObj from this tscObj
-      
+  
+  SDispatcherManager*dispatcherManager;
   SReqOrigin         from;
 } STscObj;
 
@@ -400,9 +402,10 @@ typedef struct SSqlObj {
   struct SSqlObj  *prev, *next;
   int64_t          self;
   
-  // connect alive
+
   int64_t          lastAlive;
   void *           pPrevContext;
+  bool             enableBatch;
 } SSqlObj;
 
 typedef struct SSqlStream {
@@ -503,7 +506,7 @@ void tscCloseTscObj(void *pObj);
 TAOS *taos_connect_a(char *ip, char *user, char *pass, char *db, uint16_t port, void (*fp)(void *, TAOS_RES *, int),
                      void *param, TAOS **taos);
 TAOS_RES* taos_query_h(TAOS* taos, const char *sqlstr, int64_t* res);
-TAOS_RES * taos_query_ra(TAOS *taos, const char *sqlstr, __async_cb_func_t fp, void *param);
+TAOS_RES * taos_query_ra(TAOS *taos, const char *sqlstr, __async_cb_func_t fp, void *param, bool enableBatch);
 // get taos connection unused session number
 int32_t taos_unused_session(TAOS* taos);
 
diff --git a/src/client/src/tscAsync.c b/src/client/src/tscAsync.c
index e75ce8dc0106bafa53411f585ad8b708c1c1bf99..c8ef467da4a15feebd3d76976679a3c22e8a091f 100644
--- a/src/client/src/tscAsync.c
+++ b/src/client/src/tscAsync.c
@@ -13,16 +13,19 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
+
 #include "os.h"
+#include "osAtomic.h"
+#include "tarray.h"
 #include "tutil.h"
 
+#include "qTableMeta.h"
 #include "tnote.h"
 #include "trpc.h"
+#include "tscBatchWrite.h"
 #include "tscLog.h"
 #include "tscSubquery.h"
 #include "tscUtil.h"
-#include "tsched.h"
-#include "qTableMeta.h"
 #include "tsclient.h"
 
 static void tscAsyncQueryRowsForNextVnode(void *param, TAOS_RES *tres, int numOfRows);
@@ -334,6 +337,26 @@ bool appendTagsFilter(SSqlObj* pSql) {
     return false;
   }
 
+  // check tags is blank or ''
+  char* p1 = pTscObj->tags;
+  if (strcmp(p1, "\'\'") == 0) {
+    tscDebug("TAGS 0x%" PRIx64 " tags is empty. user=%s", pSql->self, pTscObj->user);
+    return false;
+  }
+  bool blank = true;
+  while(*p1 != 0) {
+    if(*p1 != ' ') {
+      blank = false;
+      break;
+    }
+    ++p1;
+  }
+  // result
+  if(blank) {
+    tscDebug("TAGS 0x%" PRIx64 " tags is all blank. user=%s", pSql->self, pTscObj->user);
+    return false;
+  }
+
   char * p = insertTags(pSql->sqlstr, pTscObj->tags);
   if(p == NULL) {
     return false;
@@ -378,7 +401,6 @@ void doAsyncQuery(STscObj* pObj, SSqlObj* pSql, __async_cb_func_t fp, void* para
   pCmd->resColumnId = TSDB_RES_COL_ID;
 
   taosAcquireRef(tscObjRef, pSql->self);
-
   int32_t code = tsParseSql(pSql, true);
 
   if (code == TSDB_CODE_TSC_ACTION_IN_PROGRESS) {
@@ -392,18 +414,27 @@ void doAsyncQuery(STscObj* pObj, SSqlObj* pSql, __async_cb_func_t fp, void* para
     taosReleaseRef(tscObjRef, pSql->self);
     return;
   }
-
-  SQueryInfo* pQueryInfo = tscGetQueryInfo(pCmd);
+  
+  if (pObj->dispatcherManager != NULL) {
+    SAsyncBatchWriteDispatcher * dispatcher = dispatcherAcquire(pObj->dispatcherManager);
+    if (dispatcherTryDispatch(dispatcher, pSql)) {
+      taosReleaseRef(tscObjRef, pSql->self);
+      tscDebug("sql obj %p has been buffer in insert buffer", pSql);
+      return;
+    }
+  }
+  
+  SQueryInfo *pQueryInfo = tscGetQueryInfo(pCmd);
   executeQuery(pSql, pQueryInfo);
   taosReleaseRef(tscObjRef, pSql->self);
 }
 
 // TODO return the correct error code to client in tscQueueAsyncError
 void taos_query_a(TAOS *taos, const char *sqlstr, __async_cb_func_t fp, void *param) {
-  taos_query_ra(taos, sqlstr, fp, param);
+  taos_query_ra(taos, sqlstr, fp, param, tsWriteBatchSize > 0);
 }
 
-TAOS_RES * taos_query_ra(TAOS *taos, const char *sqlstr, __async_cb_func_t fp, void *param) {
+TAOS_RES * taos_query_ra(TAOS *taos, const char *sqlstr, __async_cb_func_t fp, void *param, bool enableBatch) {
   STscObj *pObj = (STscObj *)taos;
   if (pObj == NULL || pObj->signature != pObj) {
     tscError("pObj:%p is NULL or freed", pObj);
@@ -429,6 +460,8 @@ TAOS_RES * taos_query_ra(TAOS *taos, const char *sqlstr, __async_cb_func_t fp, v
     return NULL;
   }
   
+  pSql->enableBatch = enableBatch;
+  
   doAsyncQuery(pObj, pSql, fp, param, sqlstr, sqlLen);
 
   return pSql;
diff --git a/src/client/src/tscBatchMerge.c b/src/client/src/tscBatchMerge.c
new file mode 100644
index 0000000000000000000000000000000000000000..177de143f9ab26ba63c7b9db6819e137876220e0
--- /dev/null
+++ b/src/client/src/tscBatchMerge.c
@@ -0,0 +1,554 @@
+/*
+* Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+*
+* This program is free software: you can use, redistribute, and/or modify
+* it under the terms of the GNU Affero General Public License, version 3
+* or later ("AGPL"), as published by the Free Software Foundation.
+*
+* This program is distributed in the hope that it will be useful, but WITHOUT
+* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+* FITNESS FOR A PARTICULAR PURPOSE.
+*
+* You should have received a copy of the GNU Affero General Public License
+* along with this program. If not, see <http://www.gnu.org/licenses/>.
+*/
+
+#include "tscBatchMerge.h"
+
+/**
+ * A util function to compare two SName.
+ */
+static int32_t compareSName(const void *left, const void *right) {
+  if (left == right) {
+    return 0;
+  }
+  SName* x = *(SName** ) left;
+  SName* y = *(SName** ) right;
+  return memcmp(x, y, sizeof(SName));
+}
+
+/**
+ * A util function to sort SArray<SMemRow> by key.
+ */
+static int32_t compareSMemRow(const void *x, const void *y) {
+  TSKEY left = memRowKey(*(void **) x);
+  TSKEY right = memRowKey(*(void **) y);
+  if (left == right) {
+    return 0;
+  } else {
+    return left > right ? 1 : -1;
+  }
+}
+
+/**
+ * If the SSubmitBlkBuilder of pBlock->uid is present, returns it. Otherwise, build a new SSubmitBlkBuilder.
+ * 
+ * @param builder the SSubmitMsgBlocksBuilder.
+ * @param pBlock  the SSubmitBlk.
+ * @return        the SSubmitBlkBuilder (NULL means failure).
+ */
+inline static SSubmitBlkBuilder* computeIfAbsentSSubmitBlkBuilder(SSubmitMsgBlocksBuilder* builder, SSubmitBlk* pBlock) {
+  SSubmitBlkBuilder** iter = taosHashGet(builder->blockBuilders, &pBlock->uid, sizeof(pBlock->uid));
+  SSubmitBlkBuilder* blocksBuilder = NULL;
+  if (iter) {
+    return *iter;
+  }
+  
+  blocksBuilder = createSSubmitBlkBuilder(pBlock);
+  if (!blocksBuilder) {
+    return NULL;
+  }
+    
+  if (taosHashPut(builder->blockBuilders, &pBlock->uid, sizeof(pBlock->uid), &blocksBuilder, sizeof(SArray*))) {
+    destroySSubmitBlkBuilder(blocksBuilder);
+    return NULL;
+  }
+  
+  return blocksBuilder;
+}
+
+SName** buildSTableNameListBuilder(STableNameListBuilder* builder, size_t* nTables) {
+  if (!taosArrayGetSize(builder->pTableNameList)) {
+    *nTables = 0;
+    return NULL;
+  }
+
+  // sort and unique.
+  taosArraySort(builder->pTableNameList, compareSName);
+  size_t tail = 0;
+  size_t nNames = taosArrayGetSize(builder->pTableNameList);
+  for (size_t i = 1; i < nNames; ++i) {
+    SName* last = taosArrayGetP(builder->pTableNameList, tail);
+    SName* current = taosArrayGetP(builder->pTableNameList, i);
+    if (memcmp(last, current, sizeof(SName)) != 0) {
+      ++tail;
+      taosArraySet(builder->pTableNameList, tail, &current);
+    }
+  }
+
+  // build table names list.
+  SName** tableNames = calloc(tail + 1, sizeof(SName*));
+  if (!tableNames) {
+    return NULL;
+  }
+  
+  // clone data.
+  for (size_t i = 0; i <= tail; ++i) {
+    SName* clone = malloc(sizeof(SName));
+    if (!clone) {
+      goto error;
+    }
+    memcpy(clone, taosArrayGetP(builder->pTableNameList, i), sizeof(SName));
+    tableNames[i] = clone;
+  }
+
+  *nTables = tail + 1;
+  return tableNames;
+
+error:
+  for (size_t i = 0; i <= tail; ++i) {
+    if (tableNames[i]) {
+      free(tableNames[i]);
+    }
+  }
+  free(tableNames);
+  return NULL;
+}
+
+SSubmitBlkBuilder* createSSubmitBlkBuilder(SSubmitBlk* metadata) {
+  SSubmitBlkBuilder* builder = calloc(1, sizeof(SSubmitBlkBuilder));
+  if (!builder) {
+    return NULL;
+  }
+
+  builder->rows = taosArrayInit(1, sizeof(SMemRow));
+  if (!builder->rows) {
+    free(builder);
+    return NULL;
+  }
+
+  builder->metadata = calloc(1, sizeof(SSubmitBlk));
+  if (!builder->metadata) {
+    taosArrayDestroy(&builder->rows);
+    free(builder);
+    return NULL;
+  }
+  memcpy(builder->metadata, metadata, sizeof(SSubmitBlk));
+
+  return builder;
+}
+
+void destroySSubmitBlkBuilder(SSubmitBlkBuilder* builder) {
+  if (!builder) {
+    return;
+  }
+  taosArrayDestroy(&builder->rows);
+  free(builder->metadata);
+  free(builder);
+}
+
+bool appendSSubmitBlkBuilder(SSubmitBlkBuilder* builder, SSubmitBlk* pBlock) {
+  assert(pBlock->uid == builder->metadata->uid);
+  assert(pBlock->schemaLen == 0);
+  
+  // shadow copy all the SMemRow to SSubmitBlkBuilder::rows.
+  char* pRow = pBlock->data;
+  char* pEnd = pBlock->data + htonl(pBlock->dataLen);
+  while (pRow < pEnd) {
+    if (!taosArrayPush(builder->rows, &pRow)) {
+      return false;
+    }
+    pRow += memRowTLen(pRow);
+  }
+
+  return true;
+}
+
+size_t writeSSubmitBlkBuilder(SSubmitBlkBuilder* builder, SSubmitBlk* target, size_t* nRows) {
+  memcpy(target, builder->metadata, sizeof(SSubmitBlk));
+
+  // sort SSubmitBlkBuilder::rows by timestamp.
+  uint32_t dataLen = 0;
+  taosArraySort(builder->rows, compareSMemRow);
+  
+  // deep copy all the SMemRow to target.
+  size_t nMemRows = taosArrayGetSize(builder->rows);
+  for (int i = 0; i < nMemRows; ++i) {
+    char* pRow = taosArrayGetP(builder->rows, i);
+    memcpy(POINTER_SHIFT(target->data, dataLen), pRow, memRowTLen(pRow));
+    dataLen += memRowTLen(pRow);
+  }
+
+  *nRows = nMemRows;
+
+  target->schemaLen = 0;
+  target->dataLen = (int32_t) htonl(dataLen);
+  target->numOfRows = (int16_t) htons(*nRows);
+
+  return dataLen + sizeof(SSubmitBlk);
+}
+
+size_t nWriteSSubmitBlkBuilder(SSubmitBlkBuilder* builder) {
+  size_t dataLen = 0;
+  size_t nRows = taosArrayGetSize(builder->rows);
+  for (int i = 0; i < nRows; ++i) {
+    char* pRow = taosArrayGetP(builder->rows, i);
+    dataLen += memRowTLen(pRow);
+  }
+  return dataLen + sizeof(SSubmitBlk);
+}
+
+SSubmitMsgBlocksBuilder* createSSubmitMsgBuilder(int64_t vgId) {
+  SSubmitMsgBlocksBuilder* builder = calloc(1, sizeof(SSubmitMsgBlocksBuilder));
+  if (!builder) {
+    return NULL;
+  }
+  builder->vgId = vgId;
+
+  builder->blockBuilders = taosHashInit(1, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), true, false);
+  if (!builder->blockBuilders) {
+    free(builder);
+    return NULL;
+  }
+  return builder;
+}
+
+size_t nWriteSSubmitMsgBuilder(SSubmitMsgBlocksBuilder* builder) {
+  size_t nWrite = 0;
+  SSubmitBlkBuilder** iter = taosHashIterate(builder->blockBuilders, NULL);
+  while (iter) {
+    SSubmitBlkBuilder* blocksBuilder = *iter;
+    nWrite += nWriteSSubmitBlkBuilder(blocksBuilder);
+    iter = taosHashIterate(builder->blockBuilders, iter);
+  }
+  return nWrite;
+}
+
+size_t writeSSubmitMsgBlocksBuilder(SSubmitMsgBlocksBuilder* builder, SSubmitBlk* pBlocks, size_t* nRows) {
+  size_t nWrite = 0;
+  SSubmitBlkBuilder** iter = taosHashIterate(builder->blockBuilders, NULL);
+  
+  // copy all the SSubmitBlk to pBlocks.
+  while (iter) {
+    size_t nSubRows = 0;
+    SSubmitBlkBuilder* blocksBuilder = *iter;
+    SSubmitBlk* pBlock = POINTER_SHIFT(pBlocks, nWrite);
+    nWrite += writeSSubmitBlkBuilder(blocksBuilder, pBlock, &nSubRows);
+    *nRows += nSubRows;
+    iter = taosHashIterate(builder->blockBuilders, iter);
+  }
+  return nWrite;
+}
+
+size_t nBlockSSubmitMsgBlocksBuilder(SSubmitMsgBlocksBuilder* builder) {
+  return taosHashGetSize(builder->blockBuilders);
+}
+
+void destroySSubmitMsgBuilder(SSubmitMsgBlocksBuilder* builder) {
+  if (!builder) {
+    return;
+  }
+
+  SSubmitBlkBuilder** iter = taosHashIterate(builder->blockBuilders, NULL);
+  while (iter) {
+    destroySSubmitBlkBuilder(*iter);
+    iter = taosHashIterate(builder->blockBuilders, iter);
+  }
+  taosHashCleanup(builder->blockBuilders);
+  free(builder);
+}
+
+bool appendSSubmitMsgBlocks(SSubmitMsgBlocksBuilder* builder, SSubmitBlk* pBlocks, size_t nBlocks) {
+  SSubmitBlk* pBlock = pBlocks;
+  for (size_t i = 0; i < nBlocks; ++i) {
+    // not support SSubmitBlk with schema.
+    assert(pBlock->schemaLen == 0);
+    
+    // get the builder of specific table (by pBlock->uid).
+    SSubmitBlkBuilder* blocksBuilder = computeIfAbsentSSubmitBlkBuilder(builder, pBlock);
+    if (!blocksBuilder) {
+      return false;
+    }
+
+    if (!appendSSubmitBlkBuilder(blocksBuilder, pBlock)) {
+      return false;
+    }
+    
+    // go to next block.
+    size_t blockSize = sizeof (SSubmitBlk) + htonl(pBlock->dataLen);
+    pBlock = POINTER_SHIFT(pBlock, blockSize);
+  }
+  return true;
+}
+
+STableDataBlocksBuilder* createSTableDataBlocksBuilder(int64_t vgId) {
+  STableDataBlocksBuilder* builder = calloc(1, sizeof(STableDataBlocksBuilder));
+  if (!builder) {
+    return NULL;
+  }
+
+  builder->blocksBuilder = createSSubmitMsgBuilder(vgId);
+  if (!builder->blocksBuilder) {
+    free(builder);
+    return NULL;
+  }
+
+  builder->vgId = vgId;
+  builder->firstBlock = NULL;
+  return builder;
+}
+
+void destroySTableDataBlocksBuilder(STableDataBlocksBuilder* builder) {
+  if (!builder) {
+    return;
+  }
+
+  destroySSubmitMsgBuilder(builder->blocksBuilder);
+  free(builder);
+}
+
+bool appendSTableDataBlocksBuilder(STableDataBlocksBuilder* builder, STableDataBlocks* dataBlocks) {
+  // the data blocks vgId must be same with builder vgId.
+  if (!dataBlocks || dataBlocks->vgId != builder->vgId) {
+    return false;
+  }
+
+  if (!builder->firstBlock) {
+    builder->firstBlock = dataBlocks;
+  }
+
+  SSubmitBlk* pBlocks = (SSubmitBlk *)(dataBlocks->pData + dataBlocks->headerSize);
+  return appendSSubmitMsgBlocks(builder->blocksBuilder, pBlocks, dataBlocks->numOfTables);
+}
+
+STableDataBlocks* buildSTableDataBlocksBuilder(STableDataBlocksBuilder* builder, size_t* nRows) {
+  SSubmitMsgBlocksBuilder* blocksBuilder = builder->blocksBuilder;
+  STableDataBlocks *firstBlock = builder->firstBlock;
+  if (!firstBlock) {
+    return NULL;
+  }
+
+  size_t nWriteSize = nWriteSSubmitMsgBuilder(builder->blocksBuilder);
+  size_t nHeaderSize = firstBlock->headerSize;
+  size_t nAllocSize = nWriteSize + nHeaderSize;
+  
+  // allocate data blocks.
+  STableDataBlocks* dataBlocks = NULL;
+  int32_t code = tscCreateDataBlock(nAllocSize, 0, (int32_t) nHeaderSize, &firstBlock->tableName, firstBlock->pTableMeta, &dataBlocks);
+  if (code != TSDB_CODE_SUCCESS) {
+    return NULL;
+  }
+  
+  // build the header (using first block).
+  dataBlocks->size = nHeaderSize;
+  memcpy(dataBlocks->pData, firstBlock->pData, nHeaderSize);
+  
+  // build the SSubmitMsg::blocks.
+  dataBlocks->size += writeSSubmitMsgBlocksBuilder(blocksBuilder, (SSubmitBlk *) (dataBlocks->pData + nHeaderSize), nRows);
+  dataBlocks->numOfTables = (int32_t) nBlockSSubmitMsgBlocksBuilder(blocksBuilder);
+  return dataBlocks;
+}
+
+STableDataBlocksListBuilder* createSTableDataBlocksListBuilder() {
+  STableDataBlocksListBuilder* builder = calloc(1, sizeof(STableDataBlocksListBuilder));
+  if (!builder) {
+    return NULL;
+  }
+
+  builder->dataBlocksBuilders = taosHashInit(8, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), true, false);
+  if (!builder->dataBlocksBuilders) {
+    free(builder);
+    return NULL;
+  }
+
+  return builder;
+}
+
+void destroySTableDataBlocksListBuilder(STableDataBlocksListBuilder* builder) {
+  if (!builder) {
+    return;
+  }
+  
+  STableDataBlocksBuilder** iter = taosHashIterate(builder->dataBlocksBuilders, NULL);
+  while (iter) {
+    destroySTableDataBlocksBuilder(*iter);
+    iter = taosHashIterate(builder->dataBlocksBuilders, iter);
+  }
+  taosHashCleanup(builder->dataBlocksBuilders);
+  free(builder);
+}
+
+bool appendSTableDataBlocksListBuilder(STableDataBlocksListBuilder* builder, STableDataBlocks* dataBlocks) {
+  // get the data blocks builder of specific vgId.
+  STableDataBlocksBuilder** item = taosHashGet(builder->dataBlocksBuilders, &dataBlocks->vgId, sizeof(dataBlocks->vgId));
+  STableDataBlocksBuilder* blocksBuilder = NULL;
+  if (item) {
+    blocksBuilder = *item;
+  } else {
+    blocksBuilder = createSTableDataBlocksBuilder(dataBlocks->vgId);
+    if (!blocksBuilder) {
+      return false;
+    }
+
+    if (taosHashPut(builder->dataBlocksBuilders, &dataBlocks->vgId, sizeof(dataBlocks->vgId), &blocksBuilder, sizeof(STableDataBlocksBuilder*))) {
+      destroySTableDataBlocksBuilder(blocksBuilder);
+      return false;
+    }
+  }
+  
+  // append to this builder.
+  return appendSTableDataBlocksBuilder(blocksBuilder, dataBlocks);
+}
+
+SArray* buildSTableDataBlocksListBuilder(STableDataBlocksListBuilder* builder, size_t* nTables, size_t* nRows) {
+  SArray* pVnodeDataBlockList = taosArrayInit(taosHashGetSize(builder->dataBlocksBuilders), sizeof(STableDataBlocks*));
+  if (!pVnodeDataBlockList) {
+    return NULL;
+  }
+  
+  // build data blocks of each vgId.
+  STableDataBlocksBuilder** iter = taosHashIterate(builder->dataBlocksBuilders, NULL);
+  while (iter) {
+    size_t nSubRows = 0;
+    STableDataBlocksBuilder* dataBlocksBuilder = *iter;
+    STableDataBlocks* dataBlocks = buildSTableDataBlocksBuilder(dataBlocksBuilder, &nSubRows);
+    if (!dataBlocks) {
+      goto error;
+    }
+    *nTables += dataBlocks->numOfTables;
+    *nRows += nSubRows;
+
+    taosArrayPush(pVnodeDataBlockList, &dataBlocks);
+    iter = taosHashIterate(builder->dataBlocksBuilders, iter);
+  }
+  return pVnodeDataBlockList;
+
+error:
+  for (int i = 0; i < taosArrayGetSize(pVnodeDataBlockList); ++i) {
+    STableDataBlocks* dataBlocks = taosArrayGetP(pVnodeDataBlockList, i);
+    tscDestroyDataBlock(NULL, dataBlocks, false);
+  }
+  taosArrayDestroy(&pVnodeDataBlockList);
+  return NULL;
+}
+
+STableNameListBuilder* createSTableNameListBuilder() {
+  STableNameListBuilder* builder = calloc(1, sizeof(STableNameListBuilder));
+  if (!builder) {
+    return NULL;
+  }
+
+  builder->pTableNameList = taosArrayInit(1, sizeof(SName*));
+  if (!builder->pTableNameList) {
+    free(builder);
+    return NULL;
+  }
+
+  return builder;
+}
+
+void destroySTableNameListBuilder(STableNameListBuilder* builder) {
+  if (!builder) {
+    return;
+  }
+  taosArrayDestroy(&builder->pTableNameList);
+  free(builder);
+}
+
+bool insertSTableNameListBuilder(STableNameListBuilder* builder, SName* name) {
+  return taosArrayPush(builder->pTableNameList, &name);
+}
+
+int32_t tscMergeSSqlObjs(SSqlObj** polls, size_t nPolls, SSqlObj* result) {
+  // statement array is empty.
+  if (!polls || !nPolls) {
+    return TSDB_CODE_TSC_INVALID_OPERATION;
+  }
+
+  STableDataBlocksListBuilder* builder = createSTableDataBlocksListBuilder();
+  if (!builder) {
+    return TSDB_CODE_TSC_OUT_OF_MEMORY;
+  }
+
+  STableNameListBuilder* nameListBuilder = createSTableNameListBuilder();
+  if (!nameListBuilder) {
+    destroySTableDataBlocksListBuilder(builder);
+    return TSDB_CODE_TSC_OUT_OF_MEMORY;
+  }
+
+  // append the existing data blocks to builder.
+  for (size_t i = 0; i < nPolls; ++i) {
+    SSqlObj *pSql = polls[i];
+    SInsertStatementParam* pInsertParam = &pSql->cmd.insertParam;
+    if (!pInsertParam->pDataBlocks) {
+      continue;
+    }
+
+    assert(pInsertParam->payloadType == PAYLOAD_TYPE_KV);
+    assert(!pInsertParam->schemaAttached);
+
+    // append each vnode data block to the builder.
+    size_t nBlocks = taosArrayGetSize(pInsertParam->pDataBlocks);
+    for (size_t j = 0; j < nBlocks; ++j) {
+      STableDataBlocks* tableBlock = taosArrayGetP(pInsertParam->pDataBlocks, j);
+      if (!appendSTableDataBlocksListBuilder(builder, tableBlock)) {
+        destroySTableDataBlocksListBuilder(builder);
+        destroySTableNameListBuilder(nameListBuilder);
+        return TSDB_CODE_TSC_OUT_OF_MEMORY;
+      }
+
+      for (int k = 0; k < pInsertParam->numOfTables; ++k) {
+        if (!insertSTableNameListBuilder(nameListBuilder, pInsertParam->pTableNameList[k])) {
+          destroySTableDataBlocksListBuilder(builder);
+          destroySTableNameListBuilder(nameListBuilder);
+          return TSDB_CODE_TSC_OUT_OF_MEMORY;
+        }
+      }
+    }
+  }
+
+  // build the vnode data blocks.
+  size_t nBlocks = 0;
+  size_t nRows = 0;
+  SInsertStatementParam* pInsertParam = &result->cmd.insertParam;
+  SArray* pVnodeDataBlocksList = buildSTableDataBlocksListBuilder(builder, &nBlocks, &nRows);
+  if (!pVnodeDataBlocksList) {
+    destroySTableDataBlocksListBuilder(builder);
+    destroySTableNameListBuilder(nameListBuilder);
+    return TSDB_CODE_TSC_OUT_OF_MEMORY;
+  }
+
+  // build the table name list.
+  size_t nTables = 0;
+  SName** pTableNameList = buildSTableNameListBuilder(nameListBuilder, &nTables);
+  if (!pTableNameList) {
+    destroySTableDataBlocksListBuilder(builder);
+    destroySTableNameListBuilder(nameListBuilder);
+    return TSDB_CODE_TSC_OUT_OF_MEMORY;
+  }
+  
+  if (nTables != nBlocks) {
+    destroySTableDataBlocksListBuilder(builder);
+    destroySTableNameListBuilder(nameListBuilder);
+    return TSDB_CODE_TSC_INVALID_OPERATION;
+  }
+
+  // replace table name list.
+  if (pInsertParam->pTableNameList) {
+    destroyTableNameList(pInsertParam);
+  }
+  pInsertParam->pTableNameList = pTableNameList;
+  pInsertParam->numOfTables = (int32_t) nTables;
+
+  // replace vnode data blocks.
+  if (pInsertParam->pDataBlocks) {
+    tscDestroyBlockArrayList(result, pInsertParam->pDataBlocks);
+  }
+  pInsertParam->pDataBlocks = pVnodeDataBlocksList;
+  pInsertParam->numOfRows = (int32_t) nRows;
+
+  // clean up.
+  destroySTableDataBlocksListBuilder(builder);
+  destroySTableNameListBuilder(nameListBuilder);
+  return TSDB_CODE_SUCCESS;
+}
diff --git a/src/client/src/tscBatchWrite.c b/src/client/src/tscBatchWrite.c
new file mode 100644
index 0000000000000000000000000000000000000000..7c1c7b14157c54675ac5f5df060e68c3843226e5
--- /dev/null
+++ b/src/client/src/tscBatchWrite.c
@@ -0,0 +1,603 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "osAtomic.h"
+
+#include "tscBatchMerge.h"
+#include "tscBatchWrite.h"
+#include "tscLog.h"
+#include "tscSubquery.h"
+#include "tsclient.h"
+
+/**
+ * Represents the callback function and its context.
+ */
+typedef struct {
+  __async_cb_func_t fp;
+  void*             param;
+} SCallbackHandler;
+
+/**
+ * The context of `batchResultCallback`.
+ */
+typedef struct {
+  size_t           nHandlers;
+  SCallbackHandler handler[];
+} SBatchCallbackContext;
+
+/**
+ * Get the number of insertion row in the sql statement.
+ *
+ * @param pSql      the sql statement.
+ * @return int32_t  the number of insertion row.
+ */
+inline static int32_t statementGetInsertionRows(SSqlObj* pSql) { return pSql->cmd.insertParam.numOfRows; }
+
+/**
+ * Return the error result to the callback function, and release the sql object.
+ *
+ * @param pSql  the sql object.
+ * @param code  the error code of the error result.
+ */
+inline static void tscReturnsError(SSqlObj* pSql, int code) {
+  if (pSql == NULL) {
+    return;
+  }
+
+  pSql->res.code = code;
+  tscAsyncResultOnError(pSql);
+}
+
+/**
+ * Proxy function to perform sequentially insert operation.
+ *
+ * @param param     the context of `batchResultCallback`.
+ * @param tres      the result object.
+ * @param code      the error code.
+ */
+static void batchResultCallback(void* param, TAOS_RES* tres, int32_t code) {
+  SBatchCallbackContext* context = param;
+  SSqlObj*               res = tres;
+
+  // handle corner case [context == null].
+  if (context == NULL) {
+    tscError("context in `batchResultCallback` is null, which should not happen");
+    if (tres) {
+      taosReleaseRef(tscObjRef, res->self);
+    }
+    return;
+  }
+
+  // handle corner case [res == null].
+  if (res == NULL) {
+    tscError("tres in `batchResultCallback` is null, which should not happen");
+    free(context);
+    return;
+  }
+
+  // handle results.
+  tscDebug("async batch result callback, number of item: %zu", context->nHandlers);
+  for (int i = 0; i < context->nHandlers; ++i) {
+    // the result object is shared by many sql objects.
+    // therefore, we need to increase the ref count.
+    taosAcquireRef(tscObjRef, res->self);
+
+    SCallbackHandler* handler = &context->handler[i];
+    handler->fp(handler->param, res, code);
+  }
+
+  taosReleaseRef(tscObjRef, res->self);
+  free(context);
+}
+
+int32_t dispatcherBatchBuilder(SBatchRequest* pRequest, SSqlObj** batch) {
+  assert(pRequest);
+  assert(pRequest->pRequests);
+  assert(pRequest->nRequests);
+
+  // create the callback context.
+  SBatchCallbackContext* context =
+      calloc(1, sizeof(SBatchCallbackContext) + pRequest->nRequests * sizeof(SCallbackHandler));
+  if (context == NULL) {
+    return TSDB_CODE_TSC_OUT_OF_MEMORY;
+  }
+
+  tscDebug("create batch call back context: %p", context);
+
+  // initialize the callback context.
+  context->nHandlers = pRequest->nRequests;
+  for (size_t i = 0; i < pRequest->nRequests; ++i) {
+    SSqlObj* pSql = pRequest->pRequests[i];
+    context->handler[i].fp = pSql->fp;
+    context->handler[i].param = pSql->param;
+  }
+
+  // merge the statements into single one.
+  tscDebug("start to merge %zu sql objs", pRequest->nRequests);
+  SSqlObj* pFirst = pRequest->pRequests[0];
+  int32_t  code = tscMergeSSqlObjs(pRequest->pRequests, pRequest->nRequests, pFirst);
+  if (code != TSDB_CODE_SUCCESS) {
+    const char* msg = tstrerror(code);
+    tscDebug("failed to merge sql objects: %s", msg);
+    free(context);
+    return code;
+  }
+
+  pFirst->fp = batchResultCallback;
+  pFirst->param = context;
+  pFirst->fetchFp = pFirst->fp;
+  taosAcquireRef(tscObjRef, pFirst->self);
+  *batch = pFirst;
+
+  for (int i = 0; i < pRequest->nRequests; ++i) {
+    SSqlObj* pSql = pRequest->pRequests[i];
+    taosReleaseRef(tscObjRef, pSql->self);
+  }
+  return code;
+}
+
+/**
+ * Poll all the SSqlObj* in the dispatcher's buffer (No Lock). After call this function,
+ * you need to notify dispatcher->notFull by yourself.
+ *
+ * @param dispatcher    the dispatcher.
+ * @param nPolls        the number of polled SSqlObj*.
+ * @return              all the SSqlObj* in the buffer.
+ */
+inline static SBatchRequest* dispatcherPollAll(SAsyncBatchWriteDispatcher* dispatcher) {
+  if (!dispatcher->bufferSize) {
+    return NULL;
+  }
+
+  SBatchRequest* pRequest = malloc(sizeof(SBatchRequest) + sizeof(SSqlObj*) * dispatcher->bufferSize);
+  if (pRequest == NULL) {
+    tscError("failed to poll all items: out of memory");
+    return NULL;
+  }
+  
+  memcpy(pRequest->pRequests, dispatcher->buffer, sizeof(SSqlObj*) * dispatcher->bufferSize);
+  pRequest->nRequests = dispatcher->bufferSize;
+  dispatcher->currentSize = 0;
+  dispatcher->bufferSize = 0;
+  return pRequest;
+}
+
+/**
+ * Poll all the SSqlObj* in the dispatcher's buffer.
+ *
+ * @param dispatcher    the dispatcher.
+ * @return              all the SSqlObj* in the buffer.
+ */
+inline static SBatchRequest* dispatcherLockPollAll(SAsyncBatchWriteDispatcher* dispatcher) {
+  SBatchRequest* pRequest = NULL;
+  pthread_mutex_lock(&dispatcher->bufferMutex);
+  pRequest = dispatcherPollAll(dispatcher);
+  pthread_cond_broadcast(&dispatcher->notFull);
+  pthread_mutex_unlock(&dispatcher->bufferMutex);
+  return pRequest;
+}
+
+/**
+ * @brief Try to offer the SSqlObj* to the dispatcher.
+ *
+ * @param dispatcher  the async bulk write dispatcher.
+ * @param pSql        the sql object to offer.
+ * @return            return whether offer success.
+ */
+inline static bool dispatcherTryOffer(SAsyncBatchWriteDispatcher* dispatcher, SSqlObj* pSql) {
+  pthread_mutex_lock(&dispatcher->bufferMutex);
+
+  // if dispatcher is shutdown, must fail back to normal insertion.
+  // usually not happen, unless taos_query_a(...) after taos_close(...).
+  if (atomic_load_8(&dispatcher->shutdown)) {
+    pthread_mutex_unlock(&dispatcher->bufferMutex);
+    return false;
+  }
+
+  // the buffer is full.
+  while (dispatcher->currentSize >= dispatcher->batchSize) {
+    if (pthread_cond_wait(&dispatcher->notFull, &dispatcher->bufferMutex)) {
+      pthread_mutex_unlock(&dispatcher->bufferMutex);
+      return false;
+    }
+  }
+
+  dispatcher->buffer[dispatcher->bufferSize++] = pSql;
+  dispatcher->currentSize += statementGetInsertionRows(pSql);
+  tscDebug("sql obj %p has been write to insert buffer", pSql);
+
+  if (dispatcher->currentSize < dispatcher->batchSize) {
+    pthread_mutex_unlock(&dispatcher->bufferMutex);
+    return true;
+  }
+
+  // the dispatcher reaches batch size.
+  SBatchRequest* pRequest = dispatcherPollAll(dispatcher);
+  pthread_cond_broadcast(&dispatcher->notFull);
+  pthread_mutex_unlock(&dispatcher->bufferMutex);
+
+  if (pRequest) {
+    dispatcherAsyncExecute(pRequest);
+  }
+  return true;
+}
+
+void dispatcherExecute(SBatchRequest* pRequest) {
+  int32_t code = TSDB_CODE_SUCCESS;
+  // no item in the buffer (items has been taken by other threads).
+  if (!pRequest) {
+    return;
+  }
+
+  assert(pRequest->pRequests);
+  assert(pRequest->nRequests);
+
+  // merge the statements into single one.
+  SSqlObj* pSql = NULL;
+  code = dispatcherBatchBuilder(pRequest, &pSql);
+  if (code != TSDB_CODE_SUCCESS) {
+    goto _error;
+  }
+
+  tscDebug("merging %zu sql objs into %p", pRequest->nRequests, pSql);
+  tscHandleMultivnodeInsert(pSql);
+  return;
+_error:
+  tscError("send async batch sql obj failed, reason: %s", tstrerror(code));
+
+  // handling the failures.
+  for (size_t i = 0; i < pRequest->nRequests; ++i) {
+    SSqlObj* item = pRequest->pRequests[i];
+    tscReturnsError(item, code);
+  }
+}
+
+/**
+ * Get the timespec after `millis` ms
+ *
+ * @param t         the timespec.
+ * @param millis    the duration in milliseconds.
+ * @return          the timespec after `millis` ms.
+ */
+static inline void afterMillis(struct timespec* t, int32_t millis) {
+  t->tv_nsec += millis * 1000000L;
+  t->tv_sec += t->tv_nsec / 1000000000L;
+  t->tv_nsec %= 1000000000L;
+}
+
+/**
+ * Sleep until `timeout` timespec. When dispatcherShutdown(...) called, the function will return immediately.
+ *
+ * @param dispatcher the dispatcher thread to sleep.
+ * @param timeout    the timeout in CLOCK_REALTIME.
+ */
+inline static void timeoutManagerSleepUntil(SDispatcherTimeoutManager* manager, struct timespec* timeout) {
+  pthread_mutex_lock(&manager->sleepMutex);
+  while (true) {
+    // notified by dispatcherShutdown(...).
+    if (isShutdownSDispatcherTimeoutManager(manager)) {
+      break;
+    }
+    if (pthread_cond_timedwait(&manager->timeout, &manager->sleepMutex, timeout)) {
+      fflush(stdout);
+      break;
+    }
+  }
+  pthread_mutex_unlock(&manager->sleepMutex);
+}
+
+/**
+ * The thread to manage batching timeout.
+ */
+static void* timeoutManagerCallback(void* arg) {
+  SDispatcherTimeoutManager* manager = arg;
+  setThreadName("tscAsyncBackground");
+
+  while (!isShutdownSDispatcherTimeoutManager(manager)) {
+    struct timespec timeout;
+    clock_gettime(CLOCK_REALTIME, &timeout);
+    afterMillis(&timeout, manager->timeoutMs);
+
+    SBatchRequest* pRequest = dispatcherLockPollAll(manager->dispatcher);
+    if (pRequest) {
+      dispatcherAsyncExecute(pRequest);
+    }
+
+    // Similar to scheduleAtFixedRate in Java, if the execution time exceed
+    // `timeoutMs` milliseconds, then there will be no sleep.
+    timeoutManagerSleepUntil(manager, &timeout);
+  }
+  return NULL;
+}
+
+SAsyncBatchWriteDispatcher* createSAsyncBatchWriteDispatcher(STscObj* pClient, int32_t batchSize, int32_t timeoutMs) {
+  SAsyncBatchWriteDispatcher* dispatcher = calloc(1, sizeof(SAsyncBatchWriteDispatcher) + batchSize * sizeof(SSqlObj*));
+  if (!dispatcher) {
+    return NULL;
+  }
+
+  assert(pClient != NULL);
+
+  dispatcher->pClient = pClient;
+  dispatcher->currentSize = 0;
+  dispatcher->bufferSize = 0;
+  dispatcher->batchSize = batchSize;
+  atomic_store_8(&dispatcher->shutdown, false);
+
+  // init the mutex and the cond.
+  pthread_mutex_init(&dispatcher->bufferMutex, NULL);
+  pthread_cond_init(&dispatcher->notFull, NULL);
+
+  // init timeout manager.
+  dispatcher->timeoutManager = createSDispatcherTimeoutManager(dispatcher, timeoutMs);
+  if (!dispatcher->timeoutManager) {
+    pthread_mutex_destroy(&dispatcher->bufferMutex);
+    pthread_cond_destroy(&dispatcher->notFull);
+    free(dispatcher);
+    return NULL;
+  }
+
+  return dispatcher;
+}
+
+/**
+ * Shutdown the dispatcher and join the timeout thread.
+ *
+ * @param dispatcher the dispatcher.
+ */
+inline static void dispatcherShutdown(SAsyncBatchWriteDispatcher* dispatcher) {
+  atomic_store_8(&dispatcher->shutdown, true);
+  if (dispatcher->timeoutManager) {
+    shutdownSDispatcherTimeoutManager(dispatcher->timeoutManager);
+  }
+}
+
+void destroySAsyncBatchWriteDispatcher(SAsyncBatchWriteDispatcher* dispatcher) {
+  if (dispatcher == NULL) {
+    return;
+  }
+
+  dispatcherShutdown(dispatcher);
+
+  // poll and send all the statements in the buffer.
+  while (true) {
+    SBatchRequest* pRequest = dispatcherLockPollAll(dispatcher);
+    if (!pRequest) {
+      break;
+    }
+    dispatcherExecute(pRequest);
+    free(pRequest);
+  }
+  // destroy the timeout manager.
+  destroySDispatcherTimeoutManager(dispatcher->timeoutManager);
+
+  // destroy the mutex.
+  pthread_mutex_destroy(&dispatcher->bufferMutex);
+  pthread_cond_destroy(&dispatcher->notFull);
+
+  free(dispatcher);
+}
+
+bool dispatcherCanDispatch(SAsyncBatchWriteDispatcher* dispatcher, SSqlObj* pSql) {
+  if (pSql == NULL || !pSql->enableBatch) {
+    return false;
+  }
+
+  SSqlCmd*    pCmd = &pSql->cmd;
+  SQueryInfo* pQueryInfo = tscGetQueryInfo(pCmd);
+
+  // only support insert statement.
+  if (!TSDB_QUERY_HAS_TYPE(pQueryInfo->type, TSDB_QUERY_TYPE_INSERT)) {
+    return false;
+  }
+
+  SInsertStatementParam* pInsertParam = &pCmd->insertParam;
+
+  // file insert not support.
+  if (TSDB_QUERY_HAS_TYPE(pInsertParam->insertType, TSDB_QUERY_TYPE_FILE_INSERT)) {
+    return false;
+  }
+
+  // only support kv payload.
+  if (pInsertParam->payloadType != PAYLOAD_TYPE_KV) {
+    return false;
+  }
+
+  // no schema attached.
+  if (pInsertParam->schemaAttached) {
+    return false;
+  }
+
+  // too many insertion rows, fail back to normal insertion.
+  if (statementGetInsertionRows(pSql) >= dispatcher->batchSize) {
+    return false;
+  }
+
+  return true;
+}
+
+bool dispatcherTryDispatch(SAsyncBatchWriteDispatcher* dispatcher, SSqlObj* pSql) {
+  if (atomic_load_8(&dispatcher->shutdown)) {
+    return false;
+  }
+
+  // the sql object doesn't support bulk insertion.
+  if (!dispatcherCanDispatch(dispatcher, pSql)) {
+    return false;
+  }
+
+  // try to offer pSql to the buffer.
+  return dispatcherTryOffer(dispatcher, pSql);
+}
+
+/**
+ * Destroy the SAsyncBatchWriteDispatcher create by SDispatcherManager.
+ * @param arg the thread local SAsyncBatchWriteDispatcher.
+ */
+static void destroyDispatcher(void* arg) {
+  SAsyncBatchWriteDispatcher* dispatcher = arg;
+  if (!dispatcher) {
+    return;
+  }
+
+  destroySAsyncBatchWriteDispatcher(dispatcher);
+}
+
+SDispatcherManager* createDispatcherManager(STscObj* pClient, int32_t batchSize, int32_t timeoutMs,
+                                            bool isThreadLocal) {
+  SDispatcherManager* dispatcher = calloc(1, sizeof(SDispatcherManager));
+  if (!dispatcher) {
+    return NULL;
+  }
+
+  assert(pClient != NULL);
+
+  dispatcher->pClient = pClient;
+  dispatcher->batchSize = batchSize;
+  dispatcher->timeoutMs = timeoutMs;
+  dispatcher->isThreadLocal = isThreadLocal;
+
+  if (isThreadLocal) {
+    if (pthread_key_create(&dispatcher->key, destroyDispatcher)) {
+      free(dispatcher);
+      return NULL;
+    }
+  } else {
+    dispatcher->pGlobal = createSAsyncBatchWriteDispatcher(pClient, batchSize, timeoutMs);
+    if (!dispatcher->pGlobal) {
+      free(dispatcher);
+      return NULL;
+    }
+  }
+  return dispatcher;
+}
+
+SAsyncBatchWriteDispatcher* dispatcherAcquire(SDispatcherManager* manager) {
+  if (!manager->isThreadLocal) {
+    return manager->pGlobal;
+  }
+
+  SAsyncBatchWriteDispatcher* value = pthread_getspecific(manager->key);
+  if (value) {
+    return value;
+  }
+
+  value = createSAsyncBatchWriteDispatcher(manager->pClient, manager->batchSize, manager->timeoutMs);
+  if (value) {
+    pthread_setspecific(manager->key, value);
+    return value;
+  }
+
+  return NULL;
+}
+
+void destroyDispatcherManager(SDispatcherManager* manager) {
+  if (manager) {
+    if (manager->isThreadLocal) {
+      pthread_key_delete(manager->key);
+    }
+
+    if (manager->pGlobal) {
+      destroySAsyncBatchWriteDispatcher(manager->pGlobal);
+    }
+    free(manager);
+  }
+}
+
+SDispatcherTimeoutManager* createSDispatcherTimeoutManager(SAsyncBatchWriteDispatcher* dispatcher, int32_t timeoutMs) {
+  SDispatcherTimeoutManager* manager = calloc(1, sizeof(SDispatcherTimeoutManager));
+  if (!manager) {
+    return NULL;
+  }
+
+  manager->timeoutMs = timeoutMs;
+  manager->dispatcher = dispatcher;
+  atomic_store_8(&manager->shutdown, false);
+
+  pthread_mutex_init(&manager->sleepMutex, NULL);
+  pthread_cond_init(&manager->timeout, NULL);
+
+  // init background thread.
+  if (pthread_create(&manager->background, NULL, timeoutManagerCallback, manager)) {
+    pthread_mutex_destroy(&manager->sleepMutex);
+    pthread_cond_destroy(&manager->timeout);
+    free(manager);
+    return NULL;
+  }
+  return manager;
+}
+
+void destroySDispatcherTimeoutManager(SDispatcherTimeoutManager* manager) {
+  if (!manager) {
+    return;
+  }
+
+  shutdownSDispatcherTimeoutManager(manager);
+  manager->dispatcher->timeoutManager = NULL;
+
+  pthread_mutex_destroy(&manager->sleepMutex);
+  pthread_cond_destroy(&manager->timeout);
+  free(manager);
+}
+
+void shutdownSDispatcherTimeoutManager(SDispatcherTimeoutManager* manager) {
+  // mark shutdown, signal shutdown to timeout thread.
+  pthread_mutex_lock(&manager->sleepMutex);
+  atomic_store_8(&manager->shutdown, true);
+  pthread_cond_broadcast(&manager->timeout);
+  pthread_mutex_unlock(&manager->sleepMutex);
+
+  // make sure the timeout thread exit.
+  pthread_join(manager->background, NULL);
+}
+
+bool isShutdownSDispatcherTimeoutManager(SDispatcherTimeoutManager* manager) {
+  if (!manager) {
+    return true;
+  }
+  return atomic_load_8(&manager->shutdown);
+}
+
+/**
+ * The proxy function to call `dispatcherExecute`.
+ *
+ * @param pMsg the schedule message.
+ */
+static void dispatcherExecuteProxy(struct SSchedMsg* pMsg) {
+  SBatchRequest* pRequest = pMsg->ahandle;
+  if (!pRequest) {
+    return;
+  }
+  
+  pMsg->ahandle = NULL;
+  dispatcherExecute(pRequest);
+  free(pRequest);
+}
+
+void dispatcherAsyncExecute(SBatchRequest* pRequest) {
+  if (!pRequest) {
+    return;
+  }
+
+  assert(pRequest->pRequests);
+  assert(pRequest->nRequests);
+
+  SSchedMsg schedMsg = {0};
+  schedMsg.fp = dispatcherExecuteProxy;
+  schedMsg.ahandle = (void*) pRequest;
+  schedMsg.thandle = (void*) 1;
+  schedMsg.msg = 0;
+  taosScheduleTask(tscQhandle, &schedMsg);
+}
diff --git a/src/client/src/tscParseInsert.c b/src/client/src/tscParseInsert.c
index 38aee8a6787d5ae99b1d98ec0b2fadd42208ac5a..750d670d1be0f00d011f05e8ff960ef875d85519 100644
--- a/src/client/src/tscParseInsert.c
+++ b/src/client/src/tscParseInsert.c
@@ -1569,7 +1569,7 @@ int tsParseInsertSql(SSqlObj *pSql) {
   if (pInsertParam->numOfParams > 0) {
     goto _clean;
   }
-
+  
   // merge according to vgId
   if (!TSDB_QUERY_HAS_TYPE(pInsertParam->insertType, TSDB_QUERY_TYPE_STMT_INSERT) && taosHashGetSize(pInsertParam->pTableBlockHashList) > 0) {
     if ((code = tscMergeTableDataBlocks(pSql, pInsertParam, true)) != TSDB_CODE_SUCCESS) {
@@ -1577,6 +1577,7 @@ int tsParseInsertSql(SSqlObj *pSql) {
     }
   }
 
+  pCmd->insertParam.numOfRows = totalNum;
   code = TSDB_CODE_SUCCESS;
   goto _clean;
 
diff --git a/src/client/src/tscSQLParser.c b/src/client/src/tscSQLParser.c
index 8eb7f15fde6a878ff617e047826df91c9e2c1878..b7cc85908c425ed711121b91ccf0835d61a24cb9 100644
--- a/src/client/src/tscSQLParser.c
+++ b/src/client/src/tscSQLParser.c
@@ -862,7 +862,7 @@ int32_t tscValidateSqlInfo(SSqlObj* pSql, struct SSqlInfo* pInfo) {
       const char* msg3 = "name too long";
       const char* msg5 = "invalid user rights";
       const char* msg7 = "not support options";
-      const char* msg8 = "tags filter length must over 3 bytes.";
+      const char* msg8 = "tags filter string length must less than 255 bytes.";
 
       pCmd->command = pInfo->type;
 
@@ -903,15 +903,14 @@ int32_t tscValidateSqlInfo(SSqlObj* pSql, struct SSqlInfo* pInfo) {
           }
         } else if (pUser->type == TSDB_ALTER_USER_TAGS) {
           SStrToken* pTags = &pUser->tags;
-          if(pTags->n < 4)
-            return invalidOperationMsg(tscGetErrorMsgPayload(pCmd), msg8);
-       } else {
+          if (pTags->n > TSDB_TAGS_LEN - 1 ) return invalidOperationMsg(tscGetErrorMsgPayload(pCmd), msg8);
+        } else {
           return invalidOperationMsg(tscGetErrorMsgPayload(pCmd), msg7);
         }
       }
 
-      break;
-    }
+    break;
+  }
 
     case TSDB_SQL_CFG_LOCAL: {
       SMiscInfo  *pMiscInfo = pInfo->pMiscInfo;
diff --git a/src/client/src/tscSql.c b/src/client/src/tscSql.c
index 409c41fb1a2d0952090ff2ef90581e37007e3679..7dd1c71fb69b94d95d83a0a9fd782a802b74724f 100644
--- a/src/client/src/tscSql.c
+++ b/src/client/src/tscSql.c
@@ -21,6 +21,7 @@
 #include "tnote.h"
 #include "trpc.h"
 #include "tscLog.h"
+#include "tscBatchWrite.h"
 #include "tscSubquery.h"
 #include "tscUtil.h"
 #include "tsclient.h"
@@ -163,15 +164,30 @@ static SSqlObj *taosConnectImpl(const char *ip, const char *user, const char *pa
   pSql->fp        = fp;
   pSql->param     = param;
   pSql->cmd.command = TSDB_SQL_CONNECT;
-
+  
+  pObj->dispatcherManager = NULL;
+  if (tsWriteBatchSize > 1 && !tscEmbedded) {
+    pObj->dispatcherManager = createDispatcherManager(pObj, tsWriteBatchSize, tsWriteBatchTimeout, tsWriteBatchThreadLocal);
+    if (!pObj->dispatcherManager) {
+      terrno = TSDB_CODE_TSC_OUT_OF_MEMORY;
+      tscReleaseRpc(pRpcObj);
+      free(pSql);
+      free(pObj);
+      return NULL;
+    }
+  }
+  
   if (TSDB_CODE_SUCCESS != tscAllocPayload(&pSql->cmd, TSDB_DEFAULT_PAYLOAD_SIZE)) {
     terrno = TSDB_CODE_TSC_OUT_OF_MEMORY;
+    if (pObj->dispatcherManager) {
+      destroyDispatcherManager(pObj->dispatcherManager);
+    }
     tscReleaseRpc(pRpcObj);
     free(pSql);
     free(pObj);
     return NULL;
   }
-
+  
   if (taos != NULL) {
     *taos = pObj;
   }
diff --git a/src/client/src/tscStream.c b/src/client/src/tscStream.c
index 7ea24ce8290fe5c5fda2841e836d2cdf82e2bd40..ac6b7355d52543727413393f45ca09b581ceab7a 100644
--- a/src/client/src/tscStream.c
+++ b/src/client/src/tscStream.c
@@ -339,7 +339,7 @@ bool sqlBufSend(TAOS *taos, char *sqlBuf) {
   } while(++sleepCnt < 20);
 
   strcat(sqlBuf, ";");
-  taos_query_ra(taos, sqlBuf, cbSendValues, NULL);
+  taos_query_ra(taos, sqlBuf, cbSendValues, NULL, false);
   return true;
 }
 
diff --git a/src/client/src/tscSystem.c b/src/client/src/tscSystem.c
index 944b85e996db3364181d9a2ce4132e827cc3f406..50c7ceb733ed405b89b9fc8d4fe3491aa7fd89ec 100644
--- a/src/client/src/tscSystem.c
+++ b/src/client/src/tscSystem.c
@@ -14,18 +14,19 @@
  */
 
 #include "os.h"
+#include "qScript.h"
 #include "taosmsg.h"
+#include "tconfig.h"
+#include "tglobal.h"
+#include "tnote.h"
 #include "tref.h"
 #include "trpc.h"
-#include "tnote.h"
-#include "ttimer.h"
-#include "tsched.h"
+#include "tscBatchWrite.h"
 #include "tscLog.h"
+#include "tsched.h"
 #include "tsclient.h"
-#include "tglobal.h"
-#include "tconfig.h"
+#include "ttimer.h"
 #include "ttimezone.h"
-#include "qScript.h"
 
 // global, not configurable
 #define TSC_VAR_NOT_RELEASE 1
@@ -319,6 +320,49 @@ void taos_cleanup(void) {
   taosTmrCleanUp(p);
 }
 
+/**
+ * Set the option value (int32, uint16, int16, int8).
+ * @param cfg   the config.
+ * @param str   the value string.
+ * @return      whether is set or not.
+ */
+static bool taos_set_option_int(SGlobalCfg * cfg, const char* str) {
+  if (cfg->cfgStatus <= TAOS_CFG_CSTATUS_OPTION) {
+    char* p = NULL;
+    errno = 0;
+    long value = strtol(str, &p, 10);
+    
+    if (errno != 0 || p == str) {
+      tscError("failed to parse option: %s, value: %s", cfg->option, str);
+      return false;
+    }
+    
+    if ((float) value < cfg->minValue || (float) value > cfg->maxValue) {
+      tscError("failed to set option: %s, setValue: %ld, minValue: %f, maxValue: %f", cfg->option, value, cfg->minValue, cfg->maxValue);
+      return false;
+    }
+    
+    if (cfg->valType == TAOS_CFG_VTYPE_INT32) {
+      *((int32_t*) cfg->ptr) = (int32_t) value;
+    } else if (cfg->valType == TAOS_CFG_VTYPE_UINT16) {
+      *((uint16_t*) cfg->ptr) = (uint16_t) value;
+    } else if (cfg->valType == TAOS_CFG_VTYPE_INT16) {
+      *((int16_t*) cfg->ptr) = (int16_t) value;
+    } else if (cfg->valType == TAOS_CFG_VTYPE_INT8) {
+      *((int8_t*) cfg->ptr) = (int8_t) value;
+    } else {
+      tscError("failed to set option: %s, type expected %d", cfg->option, cfg->valType);
+      return false;
+    }
+    
+    cfg->cfgStatus = TAOS_CFG_CSTATUS_OPTION;
+    tscDebug("config option: %s has set to %s", cfg->option, str);
+    return true;
+  }
+  tscWarn("config option: %s, is configured by %s", cfg->option, tsCfgStatusStr[cfg->cfgStatus]);
+  return false;
+}
+
 static int taos_options_imp(TSDB_OPTION option, const char *pStr) {
   SGlobalCfg *cfg = NULL;
 
@@ -473,6 +517,27 @@ static int taos_options_imp(TSDB_OPTION option, const char *pStr) {
                 tsCfgStatusStr[cfg->cfgStatus], (char *)cfg->ptr);
       }
       break;
+      
+    case TSDB_WRITE_BATCH_SIZE: {
+      cfg = taosGetConfigOption("writeBatchSize");
+      assert(cfg != NULL);
+      taos_set_option_int(cfg, pStr);
+      break;
+    }
+      
+    case TSDB_WRITE_BATCH_TIMEOUT: {
+      cfg = taosGetConfigOption("writeBatchTimeout");
+      assert(cfg != NULL);
+      taos_set_option_int(cfg, pStr);
+      break;
+    }  
+      
+    case TSDB_WRITE_BATCH_THREAD_LOCAL: {
+      cfg = taosGetConfigOption("writeBatchThreadLocal");
+      assert(cfg != NULL);
+      taos_set_option_int(cfg, pStr);
+      break;
+    }
 
     default:
       // TODO return the correct error code to client in the format for taos_errstr()
diff --git a/src/client/src/tscUtil.c b/src/client/src/tscUtil.c
index 6e29d6ddc3facdfac745557b49f31d80a4a2cb11..c0a96c5b3ae0e5345344a1f6968ec4ba37d4a281 100644
--- a/src/client/src/tscUtil.c
+++ b/src/client/src/tscUtil.c
@@ -20,6 +20,7 @@
 #include "texpr.h"
 #include "tkey.h"
 #include "tmd5.h"
+#include "tscBatchWrite.h"
 #include "tscGlobalmerge.h"
 #include "tscLog.h"
 #include "tscProfile.h"
@@ -1841,32 +1842,36 @@ void tscDestroyBoundColumnInfo(SParsedDataColInfo* pColInfo) {
   tfree(pColInfo->colIdxInfo);
 }
 
-void tscDestroyDataBlock(SSqlObj *pSql, STableDataBlocks* pDataBlock, bool removeMeta) {
-  if (pDataBlock == NULL) {
+void destroySTableDataBlocks(STableDataBlocks* pDataBlocks) {
+  if (!pDataBlocks) {
     return;
   }
+  tfree(pDataBlocks->pData);
+  if (!pDataBlocks->cloned) {
+    tfree(pDataBlocks->params);
+    
+    // free the refcount for metermeta
+    if (pDataBlocks->pTableMeta != NULL) {
+      tfree(pDataBlocks->pTableMeta);
+    }
 
-  tfree(pDataBlock->pData);
+    tscDestroyBoundColumnInfo(&pDataBlocks->boundColumnInfo);
+  }
+  tfree(pDataBlocks);
+}
 
+void tscDestroyDataBlock(SSqlObj *pSql, STableDataBlocks* pDataBlock, bool removeMeta) {
+  if (pDataBlock == NULL) {
+    return;
+  }
+  
   if (removeMeta) {
     char name[TSDB_TABLE_FNAME_LEN] = {0};
     tNameExtractFullName(&pDataBlock->tableName, name);
-
     taosHashRemove(UTIL_GET_TABLEMETA(pSql), name, strnlen(name, TSDB_TABLE_FNAME_LEN));
   }
 
-  if (!pDataBlock->cloned) {
-    tfree(pDataBlock->params);
-
-    // free the refcount for metermeta
-    if (pDataBlock->pTableMeta != NULL) {
-      tfree(pDataBlock->pTableMeta);
-    }
-
-    tscDestroyBoundColumnInfo(&pDataBlock->boundColumnInfo);
-  }
-
-  tfree(pDataBlock);
+  destroySTableDataBlocks(pDataBlock);
 }
 
 SParamInfo* tscAddParamToDataBlock(STableDataBlocks* pDataBlock, char type, uint8_t timePrec, int16_t bytes,
@@ -1893,18 +1898,22 @@ SParamInfo* tscAddParamToDataBlock(STableDataBlocks* pDataBlock, char type, uint
   return param;
 }
 
-void*  tscDestroyBlockArrayList(SSqlObj *pSql, SArray* pDataBlockList) {
-  if (pDataBlockList == NULL) {
-    return NULL;
+void destroySTableDataBlocksList(SArray* pDataBlocks) {
+  if (!pDataBlocks) {
+    return;
   }
-
-  size_t size = taosArrayGetSize(pDataBlockList);
-  for (int32_t i = 0; i < size; i++) {
-    void* d = taosArrayGetP(pDataBlockList, i);
-    tscDestroyDataBlock(pSql, d, false);
+  size_t nBlocks = taosArrayGetSize(pDataBlocks);
+  for (size_t i = 0; i < nBlocks; ++i) {
+    STableDataBlocks * pDataBlock = taosArrayGetP(pDataBlocks, i);
+    if (pDataBlock) {
+      destroySTableDataBlocks(pDataBlock);
+    }
   }
+  taosArrayDestroy(&pDataBlocks);
+}
 
-  taosArrayDestroy(&pDataBlockList);
+void*  tscDestroyBlockArrayList(SSqlObj *pSql, SArray* pDataBlockList) {
+  destroySTableDataBlocksList(pDataBlockList);
   return NULL;
 }
 
@@ -2202,42 +2211,35 @@ static int32_t getRowExpandSize(STableMeta* pTableMeta) {
   return result;
 }
 
-static void extractTableNameList(SSqlObj *pSql, SInsertStatementParam *pInsertParam, bool freeBlockMap) {
-  pInsertParam->numOfTables = (int32_t) taosHashGetSize(pInsertParam->pTableBlockHashList);
-  if (pInsertParam->pTableNameList == NULL) {
-    pInsertParam->pTableNameList = malloc(pInsertParam->numOfTables * POINTER_BYTES);
-  }
-
-  STableDataBlocks **p1 = taosHashIterate(pInsertParam->pTableBlockHashList, NULL);
-  int32_t i = 0;
-  while(p1) {
-    STableDataBlocks* pBlocks = *p1;
-    //tfree(pInsertParam->pTableNameList[i]);
-
-    pInsertParam->pTableNameList[i++] = tNameDup(&pBlocks->tableName);
-    p1 = taosHashIterate(pInsertParam->pTableBlockHashList, p1);
-  }
-
-  if (freeBlockMap) {
-    pInsertParam->pTableBlockHashList = tscDestroyBlockHashTable(pSql, pInsertParam->pTableBlockHashList, false);
-  }
-}
-
 int32_t tscMergeTableDataBlocks(SSqlObj *pSql, SInsertStatementParam *pInsertParam, bool freeBlockMap) {
   const int INSERT_HEAD_SIZE = sizeof(SMsgDesc) + sizeof(SSubmitMsg);
   int       code = 0;
   bool      isRawPayload = IS_RAW_PAYLOAD(pInsertParam->payloadType);
-  void*     pVnodeDataBlockHashList = taosHashInit(128, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), true, false);
+  size_t    initialSize = taosHashGetSize(pInsertParam->pTableBlockHashList);
+  initialSize = initialSize > 128 ? 128 : initialSize;
+  
+  void*     pVnodeDataBlockHashList = taosHashInit(initialSize, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), true, false);
   SArray*   pVnodeDataBlockList = taosArrayInit(8, POINTER_BYTES);
-
-  STableDataBlocks** p = taosHashIterate(pInsertParam->pTableBlockHashList, NULL);
-
-  STableDataBlocks* pOneTableBlock = *p;
-
+  
+  // alloc table name list.
+  size_t numOfTables = taosHashGetSize(pInsertParam->pTableBlockHashList);
+  if (pInsertParam->pTableNameList) {
+    destroyTableNameList(pInsertParam);
+  }
+  pInsertParam->pTableNameList = calloc(numOfTables, sizeof(SName*));
+  pInsertParam->numOfTables = (int32_t) numOfTables;
+  
+  size_t tail = 0;
   SBlockKeyInfo blkKeyInfo = {0};  // share by pOneTableBlock
-
-  while(pOneTableBlock) {
+  STableDataBlocks** iter = taosHashIterate(pInsertParam->pTableBlockHashList, NULL);
+  while (iter) {
+    STableDataBlocks* pOneTableBlock = *iter;
     SSubmitBlk* pBlocks = (SSubmitBlk*) pOneTableBlock->pData;
+    iter = taosHashIterate(pInsertParam->pTableBlockHashList, iter);
+    
+    // extract table name list.
+    pInsertParam->pTableNameList[tail++] = tNameDup(&pOneTableBlock->tableName);
+    
     if (pBlocks->numOfRows > 0) {
       // the maximum expanded size in byte when a row-wise data is converted to SDataRow format
       int32_t           expandSize = isRawPayload ? getRowExpandSize(pOneTableBlock->pTableMeta) : 0;
@@ -2320,19 +2322,19 @@ int32_t tscMergeTableDataBlocks(SSqlObj *pSql, SInsertStatementParam *pInsertPar
       dataBuf->numOfTables += 1;
 
       pBlocks->numOfRows = 0;
-    }else {
+    } else {
       tscDebug("0x%"PRIx64" table %s data block is empty", pInsertParam->objectId, pOneTableBlock->tableName.tname);
     }
-
-    p = taosHashIterate(pInsertParam->pTableBlockHashList, p);
-    if (p == NULL) {
-      break;
+    
+    if (freeBlockMap) {
+      tscDestroyDataBlock(pSql, pOneTableBlock, false);
     }
-
-    pOneTableBlock = *p;
   }
-
-  extractTableNameList(pSql, pInsertParam, freeBlockMap);
+  
+  if (freeBlockMap) {
+    taosHashCleanup(pInsertParam->pTableBlockHashList);
+    pInsertParam->pTableBlockHashList = NULL;
+  }
 
   // free the table data blocks;
   pInsertParam->pDataBlocks = pVnodeDataBlockList;
@@ -2353,6 +2355,9 @@ void tscCloseTscObj(void *param) {
   tscReleaseRpc(pObj->pRpcObj);
   pthread_mutex_destroy(&pObj->mutex);
   tscReleaseClusterInfo(pObj->clusterId);
+  
+  destroyDispatcherManager(pObj->dispatcherManager);
+  pObj->dispatcherManager = NULL;
 
   tfree(pObj);
 }
diff --git a/src/common/inc/tglobal.h b/src/common/inc/tglobal.h
index 20153e76ea4af28996f32778fadf9191e5ccf9d9..bcc0d68f95cbd9ec884df5e87b3ec10bd5463883 100644
--- a/src/common/inc/tglobal.h
+++ b/src/common/inc/tglobal.h
@@ -92,6 +92,9 @@ extern int32_t tsRetryStreamCompDelay;
 extern float   tsStreamComputDelayRatio;  // the delayed computing ration of the whole time window
 extern int32_t tsProjectExecInterval;
 extern int64_t tsMaxRetentWindow;
+extern bool    tsWriteBatchThreadLocal;
+extern int32_t tsWriteBatchSize;
+extern int32_t tsWriteBatchTimeout;
 
 // db parameters in client
 extern int32_t tsCacheBlockSize;
diff --git a/src/common/src/tglobal.c b/src/common/src/tglobal.c
index 57b77d32b09adf99e159405652ffd382f1cf6eae..30711dfd3997cd8f6fd10dd7af8b147b137ac0c0 100644
--- a/src/common/src/tglobal.c
+++ b/src/common/src/tglobal.c
@@ -127,6 +127,11 @@ int8_t tsSortWhenGroupBy = 1;
 int32_t tsProjectExecInterval = 10000;   // every 10sec, the projection will be executed once
 int64_t tsMaxRetentWindow = 24 * 3600L;  // maximum time window tolerance
 
+// The tsc async write batching feature (using ABWD).
+bool    tsWriteBatchThreadLocal = false;    // if thread local enable, each thread will allocate a dispatcher.
+int32_t tsWriteBatchSize = 0;               // suggest: 64 - 512, default 0, 0 means disable batching.
+int32_t tsWriteBatchTimeout = 10;           // suggest: 2 - 100 (unit: milliseconds)
+
 // the maximum allowed query buffer size during query processing for each data node.
 // -1 no limit (default)
 // 0  no query allowed, queries are disabled
@@ -1874,6 +1879,37 @@ static void doInitGlobalConfig(void) {
   cfg.ptrLength = 0;
   cfg.unitType = TAOS_CFG_UTYPE_NONE;
   taosInitConfigOption(cfg);
+  
+  cfg.option = "writeBatchThreadLocal";
+  cfg.ptr = &tsWriteBatchThreadLocal;
+  cfg.valType = TAOS_CFG_VTYPE_INT8;
+  cfg.cfgType = TSDB_CFG_CTYPE_B_CONFIG | TSDB_CFG_CTYPE_B_SHOW | TSDB_CFG_CTYPE_B_CLIENT;
+  cfg.minValue = 0;
+  cfg.maxValue = 1;
+  cfg.ptrLength = 0;
+  cfg.unitType = TAOS_CFG_UTYPE_NONE;
+  taosInitConfigOption(cfg);
+
+  cfg.option = "writeBatchSize";
+  cfg.ptr = &tsWriteBatchSize;
+  cfg.valType = TAOS_CFG_VTYPE_INT32;
+  cfg.cfgType = TSDB_CFG_CTYPE_B_CONFIG | TSDB_CFG_CTYPE_B_SHOW | TSDB_CFG_CTYPE_B_CLIENT;
+  cfg.minValue = 0;
+  cfg.maxValue = 4096;
+  cfg.ptrLength = 0;
+  cfg.unitType = TAOS_CFG_UTYPE_NONE;
+  taosInitConfigOption(cfg);
+
+  cfg.option = "writeBatchTimeout";
+  cfg.ptr = &tsWriteBatchTimeout;
+  cfg.valType = TAOS_CFG_VTYPE_INT32;
+  cfg.cfgType = TSDB_CFG_CTYPE_B_CONFIG | TSDB_CFG_CTYPE_B_SHOW | TSDB_CFG_CTYPE_B_CLIENT;
+  cfg.minValue = 1;
+  cfg.maxValue = 2048;
+  cfg.ptrLength = 0;
+  cfg.unitType = TAOS_CFG_UTYPE_NONE;
+  taosInitConfigOption(cfg);
+
   assert(tsGlobalConfigNum == TSDB_CFG_MAX_NUM);
 #else
   // if TD_TSZ macro define, have 5 count configs, so must add 5
diff --git a/src/inc/taos.h b/src/inc/taos.h
index b2e2e9f3c625e05c5ebbedec6d0c84b31a3a9259..efb3ce25d678f7af83954c15e0625f697315c714 100644
--- a/src/inc/taos.h
+++ b/src/inc/taos.h
@@ -60,7 +60,10 @@ typedef enum {
   TSDB_OPTION_TIMEZONE,
   TSDB_OPTION_CONFIGDIR,
   TSDB_OPTION_SHELL_ACTIVITY_TIMER,
-  TSDB_MAX_OPTIONS
+  TSDB_MAX_OPTIONS,
+  TSDB_WRITE_BATCH_SIZE,
+  TSDB_WRITE_BATCH_TIMEOUT,
+  TSDB_WRITE_BATCH_THREAD_LOCAL
 } TSDB_OPTION;
 
 typedef struct taosField {
diff --git a/src/kit/shell/src/shellEngine.c b/src/kit/shell/src/shellEngine.c
index ecd7d2d31ecf77a5c9e41b24c99b8e68dc7f880a..1bdcc532943318f5fe368b3832145ad7e06f6a56 100644
--- a/src/kit/shell/src/shellEngine.c
+++ b/src/kit/shell/src/shellEngine.c
@@ -377,7 +377,7 @@ void shellRunCommandOnServer(TAOS *con, char command[]) {
   } else {
     int num_rows_affacted = taos_affected_rows(pSql);
     et = taosGetTimestampUs();
-    printf("Query OK, %d of %d row(s) in database (%.6fs)\n", num_rows_affacted, num_rows_affacted, (et - st) / 1E6);
+    printf("Query OK, %d row(s) affected in set (%.6fs)\n", num_rows_affacted, (et - st) / 1E6);
 
 #ifndef WINDOWS
     // call auto tab
diff --git a/src/mnode/src/mnodeUser.c b/src/mnode/src/mnodeUser.c
index f866da96cdd0b3d8b1809e0eeb53b6e41cfc067d..85f20b6d24990c9229fb47e5744d232b96508d8b 100644
--- a/src/mnode/src/mnodeUser.c
+++ b/src/mnode/src/mnodeUser.c
@@ -97,7 +97,7 @@ static int32_t mnodeUserActionDecode(SSdbRow *pRow) {
   SUserObj *pUser = (SUserObj *)calloc(1, sizeof(SUserObj));
   if (pUser == NULL) return TSDB_CODE_MND_OUT_OF_MEMORY;
 
-  memcpy(pUser, pRow->rowData, tsUserUpdateSize);
+  memcpy(pUser, pRow->rowData, pRow->rowSize);
   pRow->pObj = pUser;
   return TSDB_CODE_SUCCESS;
 }
diff --git a/src/query/src/qExecutor.c b/src/query/src/qExecutor.c
index c352c9c47a99fa27a5942081b9735bfa3878c067..d3e38cf365272781ecd3b7727d1f9809f50897af 100644
--- a/src/query/src/qExecutor.c
+++ b/src/query/src/qExecutor.c
@@ -318,10 +318,10 @@ static int32_t getColDataFromGroupRes(SGroupResInfo* pGroupResInfo, SQueryRuntim
   for (int32_t i = 0; i < numRows; ++i) {
     SResultRow* row = taosArrayGetP(pGroupResInfo->pRows, i);
     row->groupIndex = i;
-    tFilePage*  page = getResBufPage(pRuntimeEnv->pResultBuf, row->pageId);
-    int32_t     rowsToCopy = 1;
-    char*       out = colData + numOfResult * colBytes;
-    char*       in = getPosInResultPage(pRuntimeEnv->pQueryAttr, page, row->offset, colDataOffset);
+    tFilePage* page = getResBufPage(pRuntimeEnv->pResultBuf, row->pageId);
+    int32_t    rowsToCopy = 1;
+    char*      out = colData + numOfResult * colBytes;
+    char*      in = getPosInResultPage(pRuntimeEnv->pQueryAttr, page, row->offset, colDataOffset);
     memcpy(out, in, colBytes * rowsToCopy);
     numOfResult += rowsToCopy;
   }
@@ -2837,6 +2837,10 @@ static bool isCachedLastQuery(SQueryAttr* pQueryAttr) {
     return false;
   }
 
+  if (pQueryAttr->pFilters != NULL || pQueryAttr->pFilterInfo != NULL) {
+    return false;
+  }
+
   return true;
 }
 
@@ -3165,7 +3169,7 @@ static bool overlapWithTimeWindow(SQueryAttr* pQueryAttr, SDataBlockInfo* pBlock
       return true;
     }
 
-    while(w.skey < pBlockInfo->window.ekey) {
+    while (w.skey < pBlockInfo->window.ekey) {
       // add one slding
       if (pQueryAttr->interval.slidingUnit == 'n' || pQueryAttr->interval.slidingUnit == 'y')
         ekey = taosTimeAdd(ekey, pQueryAttr->interval.sliding, pQueryAttr->interval.slidingUnit, pQueryAttr->precision);
@@ -3180,7 +3184,7 @@ static bool overlapWithTimeWindow(SQueryAttr* pQueryAttr, SDataBlockInfo* pBlock
       getAlignQueryTimeWindow(pQueryAttr, ekey, sk, ek, &w);
     }
 
-    while(1) {
+    while (1) {
       if (w.ekey < pBlockInfo->window.skey) {
         break;
       }
@@ -6339,7 +6343,7 @@ SOperatorInfo* createOrderOperatorInfo(SQueryRuntimeEnv* pRuntimeEnv, SOperatorI
       SColumnInfoData col = {{0}};
       col.info.colId = pExpr[i].base.colInfo.colId;
       col.info.bytes = pExpr[i].base.resBytes;
-      col.info.type  = pExpr[i].base.resType;
+      col.info.type = pExpr[i].base.resType;
       taosArrayPush(pDataBlock->pDataBlock, &col);
 
       if (!found && col.info.colId == pOrderVal->orderColId) {
@@ -6874,18 +6878,18 @@ static bool doEveryInterpolation(SOperatorInfo* pOperatorInfo, SSDataBlock* pBlo
       break;
     }
   }
-  
+
   if (pCtx == NULL) {
     goto group_finished_exit;
   }
-  
+
   TSKEY* tsCols = NULL;
   if (pBlock && pBlock->pDataBlock != NULL) {
     SColumnInfoData* pColDataInfo = taosArrayGet(pBlock->pDataBlock, 0);
     tsCols = (int64_t*)pColDataInfo->pData;
     assert(tsCols[0] == pBlock->info.window.skey && tsCols[pBlock->info.rows - 1] == pBlock->info.window.ekey);
   }
-  
+
   if (pCtx->startTs == INT64_MIN) {
     if (pQueryAttr->range.skey == INT64_MIN) {
       if (NULL == tsCols) {
@@ -6984,7 +6988,7 @@ static bool doEveryInterpolation(SOperatorInfo* pOperatorInfo, SSDataBlock* pBlo
     return false;
   }
   int32_t nRows = pBlock ? pBlock->info.rows : 0;
-  int32_t startPos = binarySearchForKey((char*) tsCols, nRows, pCtx->startTs, pQueryAttr->order.order);
+  int32_t startPos = binarySearchForKey((char*)tsCols, nRows, pCtx->startTs, pQueryAttr->order.order);
 
   if (ascQuery && pQueryAttr->fillType != TSDB_FILL_NEXT && pCtx->start.key == INT64_MIN) {
     if (startPos < 0) {
@@ -8522,18 +8526,18 @@ SOperatorInfo* createSLimitOperatorInfo(SQueryRuntimeEnv* pRuntimeEnv, SOperator
     SColIndex* idx = taosArrayGet(pInfo->orderColumnList, i);
     offset += pExpr[idx->colIndex].base.resBytes;
   }
-  
+
   SOperatorInfo* pOperator = calloc(1, sizeof(SOperatorInfo));
   if (pOperator == NULL) {
     goto _clean;
   }
-  
+
   pInfo->pRes = createOutputBuf(pExpr, numOfOutput, pRuntimeEnv->resultInfo.capacity);
   if (pInfo->pRes == NULL) {
     tfree(pOperator);
     goto _clean;
   }
-  
+
   pOperator->name = "SLimitOperator";
   pOperator->operatorType = OP_SLimit;
   pOperator->blockingOptr = false;
@@ -10171,7 +10175,8 @@ SQInfo* createQInfoImpl(SQueryTableMsg* pQueryMsg, SGroupbyExpr* pGroupbyExpr, S
     pTableqinfo->pGroupList = taosArrayInit(numOfGroups, POINTER_BYTES);
     pTableqinfo->numOfTables = pTableGroupInfo->numOfTables;
 
-    pTableqinfo->map = taosHashInit(pTableGroupInfo->numOfTables, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), true, HASH_ENTRY_LOCK);
+    pTableqinfo->map = taosHashInit(pTableGroupInfo->numOfTables, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), true,
+                                    HASH_ENTRY_LOCK);
   }
 
   pQInfo->pBuf = calloc(pTableGroupInfo->numOfTables, sizeof(STableQueryInfo));
@@ -10373,7 +10378,7 @@ static void doDestroyTableQueryInfo(STableGroupInfo* pTableqinfoGroupInfo) {
 
   taosArrayDestroy(&pTableqinfoGroupInfo->pGroupList);
 
-  SHashObj *pmap = pTableqinfoGroupInfo->map;
+  SHashObj* pmap = pTableqinfoGroupInfo->map;
   if (pmap == atomic_val_compare_exchange_ptr(&pTableqinfoGroupInfo->map, pmap, NULL)) {
     taosHashCleanup(pmap);
   }
diff --git a/src/tsdb/src/tsdbCommit.c b/src/tsdb/src/tsdbCommit.c
index 3abc3e9acc6c8f6e909d4d6ef5f043dc2ee3e156..9358530581a3a755b1765e6e1571bf06a428b074 100644
--- a/src/tsdb/src/tsdbCommit.c
+++ b/src/tsdb/src/tsdbCommit.c
@@ -1080,7 +1080,7 @@ int tsdbWriteBlockImpl(STsdbRepo *pRepo, STable *pTable, SDFile *pDFile, SDFile
   int         rowsToWrite = pDataCols->numOfRows;
 
   ASSERT(rowsToWrite > 0 && rowsToWrite <= pCfg->maxRowsPerFileBlock);
-  ASSERT((!isLast) || rowsToWrite < pCfg->minRowsPerFileBlock);
+  // ASSERT((!isLast) || rowsToWrite < pCfg->minRowsPerFileBlock);
 
   // Make buffer space
   if (tsdbMakeRoom(ppBuf, tsdbBlockStatisSize(pDataCols->numOfCols, SBlockVerLatest)) < 0) {
diff --git a/src/util/inc/tconfig.h b/src/util/inc/tconfig.h
index f207b9b5c4dd270cc4364e548449edcbee43ae4a..d6f4c79ddd8c0d22149eeead5ef23887aae69b04 100644
--- a/src/util/inc/tconfig.h
+++ b/src/util/inc/tconfig.h
@@ -20,7 +20,7 @@
 extern "C" {
 #endif
 
-#define TSDB_CFG_MAX_NUM    138
+#define TSDB_CFG_MAX_NUM    141
 #define TSDB_CFG_PRINT_LEN  23
 #define TSDB_CFG_OPTION_LEN 24
 #define TSDB_CFG_VALUE_LEN  41
diff --git a/src/util/src/tutil.c b/src/util/src/tutil.c
index 7e08178c42cbb7e34a0aec340352916d874a7094..649f8fa5e9697eac76f4a97eb44d5b18e06924ec 100644
--- a/src/util/src/tutil.c
+++ b/src/util/src/tutil.c
@@ -225,66 +225,59 @@ char *tstrstr(char *src, char *dst, bool ignoreInEsc) {
 }
 
 char* strtolower(char *dst, const char *src) {
-  int esc = 0;
-  char quote = 0, *p = dst, c;
-
-  assert(dst != NULL);
-
-  for (c = *src++; c; c = *src++) {
-    if (esc) {
-      esc = 0;
-    } else if (quote) {
-      if (c == '\\') {
-        esc = 1;
-      } else if (c == quote) {
-        quote = 0;
+  if (src == NULL || dst == NULL) {
+    return dst;
+  }
+  char* const ret = dst;
+  while (*src) {
+    const char ch = *(src++);
+    *(dst++) = (ch >= 'A' && ch <= 'Z') ? ch - 'A' + 'a' : ch;
+         
+    if (ch == '\'' || ch == '"') {
+      char prev = ch;
+      while (*src) {
+        const char next = *(src++);
+        *(dst++) = next;
+        if (prev != '\\' && next == ch) break;
+        prev = next;
+      }
+    } else if (ch == '`') {
+      while (*src) {
+        const char next = *(src++);
+        *(dst++) = next;
+        if (next == ch) break;
       }
-    } else if (c >= 'A' && c <= 'Z') {
-      c -= 'A' - 'a';
-    } else if (c == '\'' || c == '"') {
-      quote = c;
     }
-    *p++ = c;
   }
-
-  *p = 0;
-  return dst;
+  *(dst) = 0;
+  return ret;
 }
 
 char* strntolower(char *dst, const char *src, int32_t n) {
-  int esc = 0, inEsc = 0;
-  char quote = 0, *p = dst, c;
-
   assert(dst != NULL);
-  if (n == 0) {
-    *p = 0;
-    return dst;
-  }
-  for (c = *src++; n-- > 0; c = *src++) {
-    if (esc) {
-      esc = 0;
-    } else if (quote) {
-      if (c == '\\') {
-        esc = 1;
-      } else if (c == quote) {
-        quote = 0;
+  char* const end = dst + n;
+  while (dst != end) {
+    const char ch = *(src++);
+    *(dst++) = (ch >= 'A' && ch <= 'Z') ? ch - 'A' + 'a' : ch;
+    
+    if (ch == '\'' || ch == '"') {
+      char prev = ch;
+      while (dst != end) {
+        const char next = *(src++);
+        *(dst++) = next;
+        if (prev != '\\' && next == ch) break;
+        prev = next;
       }
-    } else if (inEsc) {
-      if (c == '`') {
-        inEsc = 0;
+    } else if (ch == '`') {
+      while (dst != end) {
+        const char next = *(src++);
+        *(dst++) = next;
+        if (next == ch) break;
       }
-    } else if (c >= 'A' && c <= 'Z') {
-      c -= 'A' - 'a';
-    } else if (inEsc == 0 && (c == '\'' || c == '"')) {
-      quote = c;
-    } else if (c == '`' && quote == 0) {
-      inEsc = 1;
     }
-    *p++ = c;
   }
-
-  *p = 0;
-  return dst;
+  *(dst) = 0;
+  return dst - n;
 }
 
 char* strntolower_s(char *dst, const char *src, int32_t n) {