diff --git a/source/dnode/vnode/src/inc/tsdb.h b/source/dnode/vnode/src/inc/tsdb.h
index b698909af1268bf5e6c216ab0324b45a7479e9ca..16afa9f596f50a37a2ddfab4c0a3619148763d05 100644
--- a/source/dnode/vnode/src/inc/tsdb.h
+++ b/source/dnode/vnode/src/inc/tsdb.h
@@ -348,6 +348,7 @@ typedef struct {
   rocksdb_options_t      *options;
   rocksdb_writeoptions_t *writeoptions;
   rocksdb_readoptions_t  *readoptions;
+  rocksdb_writebatch_t   *writebatch;
   TdThreadMutex           rMutex;
 } SRocksCache;
 
@@ -805,7 +806,7 @@ typedef struct {
 
 int32_t tsdbOpenCache(STsdb *pTsdb);
 void    tsdbCloseCache(STsdb *pTsdb);
-int32_t tsdbCacheUpdate(STsdb *pTsdb, tb_uid_t uid, TSDBROW *row);
+int32_t tsdbCacheUpdate(STsdb *pTsdb, tb_uid_t suid, tb_uid_t uid, TSDBROW *row);
 int32_t tsdbCacheInsertLast(SLRUCache *pCache, tb_uid_t uid, TSDBROW *row, STsdb *pTsdb);
 int32_t tsdbCacheInsertLastrow(SLRUCache *pCache, STsdb *pTsdb, tb_uid_t uid, TSDBROW *row, bool dup);
 int32_t tsdbCacheGetLastH(SLRUCache *pCache, tb_uid_t uid, SCacheRowsReader *pr, LRUHandle **h);
diff --git a/source/dnode/vnode/src/tsdb/tsdbCache.c b/source/dnode/vnode/src/tsdb/tsdbCache.c
index 171a63aa33685f7bb98cb77bf5eaabe5faf244fb..393e8ffdf5cf7cbfe33e6dde31ed246f33f09c4b 100644
--- a/source/dnode/vnode/src/tsdb/tsdbCache.c
+++ b/source/dnode/vnode/src/tsdb/tsdbCache.c
@@ -43,6 +43,8 @@ static void tsdbCloseBICache(STsdb *pTsdb) {
   }
 }
 
+#define ROCKS_KEY_LEN 64
+
 static void tsdbGetRocksPath(STsdb *pTsdb, char *path) {
   SVnode *pVnode = pTsdb->pVnode;
   if (pVnode->pTfs) {
@@ -92,8 +94,11 @@ static int32_t tsdbOpenRocksCache(STsdb *pTsdb) {
     goto _err3;
   }
 
+  rocksdb_writebatch_t *writebatch = rocksdb_writebatch_create();
+
   taosThreadMutexInit(&pTsdb->rCache.rMutex, NULL);
 
+  pTsdb->rCache.writebatch = writebatch;
   pTsdb->rCache.options = options;
   pTsdb->rCache.writeoptions = writeoptions;
   pTsdb->rCache.readoptions = readoptions;
@@ -113,37 +118,167 @@ _err:
 
 static void tsdbCloseRocksCache(STsdb *pTsdb) {
   rocksdb_close(pTsdb->rCache.db);
+  rocksdb_writebatch_destroy(pTsdb->rCache.writebatch);
   rocksdb_readoptions_destroy(pTsdb->rCache.readoptions);
   rocksdb_writeoptions_destroy(pTsdb->rCache.writeoptions);
   rocksdb_options_destroy(pTsdb->rCache.options);
 }
 
-int32_t tsdbCacheUpdate(STsdb *pTsdb, tb_uid_t uid, TSDBROW *pRow) {
+SLastCol *tsdbCacheDeserialize(char const *value) {
+  SLastCol *pLastCol = (SLastCol *)value;
+  SColVal  *pColVal = &pLastCol->colVal;
+  if (IS_VAR_DATA_TYPE(pColVal->type)) {
+    pColVal->value.pData = (char *)value + sizeof(*pColVal);
+  }
+
+  return pLastCol;
+}
+
+void tsdbCacheSerialize(SLastCol *pLastCol, char **value, size_t *size) {
+  SColVal *pColVal = &pLastCol->colVal;
+  size_t   length = sizeof(*pLastCol);
+  if (IS_VAR_DATA_TYPE(pColVal->type)) {
+    length += pColVal->value.nData;
+  }
+  *value = taosMemoryMalloc(length);
+
+  *(SLastCol *)(*value) = *pLastCol;
+  if (IS_VAR_DATA_TYPE(pColVal->type)) {
+    uint8_t *pVal = pColVal->value.pData;
+    pColVal->value.pData = *value + sizeof(*pLastCol);
+    if (pColVal->value.nData) {
+      memcpy(pColVal->value.pData, pVal, pColVal->value.nData);
+    }
+  }
+  *size = length;
+}
+
+int32_t tsdbCacheUpdate(STsdb *pTsdb, tb_uid_t suid, tb_uid_t uid, TSDBROW *pRow) {
   int32_t code = 0;
 
+  // 1, fetch schema
+  STSchema *pTSchema = NULL;
+  int32_t   sver = TSDBROW_SVERSION(pRow);
+  code = metaGetTbTSchemaEx(pTsdb->pVnode->pMeta, suid, uid, sver, &pTSchema);
+  if (code != TSDB_CODE_SUCCESS) {
+    terrno = code;
+    return -1;
+  }
+
+  // 2, iterate col values into array
+  SArray *aColVal = taosArrayInit(32, sizeof(SColVal));
+
   STSDBRowIter iter = {0};
   tsdbRowIterOpen(&iter, pRow, pTSchema);
 
   for (SColVal *pColVal = tsdbRowIterNext(&iter); pColVal; pColVal = tsdbRowIterNext(&iter)) {
-    SColVal *pRColVal = tsdbCacheGetRColVal(pTsdb);
-    if (pRColVal) {
-      // merge pColVal with pRColVal
+    if (IS_VAR_DATA_TYPE(pColVal->type)) {
+      uint8_t *pVal = pColVal->value.pData;
+
+      pColVal->value.pData = NULL;
+      code = tRealloc(&pColVal->value.pData, pColVal->value.nData);
+      if (code) {
+        code = TSDB_CODE_OUT_OF_MEMORY;
+        goto _exit;
+      }
+
+      if (pColVal->value.nData) {
+        memcpy(pColVal->value.pData, pVal, pColVal->value.nData);
+      }
     }
 
-    tsdbCachePutRColVal(pColVal);
+    taosArrayPush(aColVal, pColVal);
   }
 
   tsdbRowClose(&iter);
 
-  char  *err = NULL;
-  char   buf[256] = {0};
-  size_t vallen = 0;
-  char  *val = rocksdb_get(pTsdb->rCache.db, pTsdb->rCache.readoptions, "key", 3, &vallen, &err);
-  if (val) {
-  } else {
+  // 3, build keys & multi get from rocks
+  int     max_key_len = 0;
+  int     num_keys = TARRAY_SIZE(aColVal);
+  char  **keys_list = taosMemoryCalloc(num_keys * 2, sizeof(char *));
+  size_t *keys_list_sizes = taosMemoryCalloc(num_keys * 2, sizeof(size_t));
+  for (int i = 0; i < num_keys; ++i) {
+    SColVal *pColVal = (SColVal *)taosArrayGet(aColVal, i);
+
+    char *keys = taosMemoryCalloc(2, ROCKS_KEY_LEN);
+    int   last_key_len = snprintf(keys, ROCKS_KEY_LEN, "%" PRIi64 ":%" PRIi16 ":last", uid, pColVal->cid);
+    if (last_key_len >= ROCKS_KEY_LEN) {
+      tsdbError("vgId:%d, %s failed at line %d since %s", TD_VID(pTsdb->pVnode), __func__, __LINE__, tstrerror(code));
+    }
+    int lr_key_len =
+        snprintf(keys + ROCKS_KEY_LEN, ROCKS_KEY_LEN, "%" PRIi64 ":%" PRIi16 ":last_row", uid, pColVal->cid);
+    if (lr_key_len >= ROCKS_KEY_LEN) {
+      tsdbError("vgId:%d, %s failed at line %d since %s", TD_VID(pTsdb->pVnode), __func__, __LINE__, tstrerror(code));
+    }
+    keys_list[i] = keys;
+    keys_list[num_keys + i] = keys + ROCKS_KEY_LEN;
+    keys_list_sizes[i] = last_key_len;
+    keys_list_sizes[num_keys + i] = lr_key_len;
+  }
+  char  **values_list = taosMemoryCalloc(num_keys * 2, sizeof(char *));
+  size_t *values_list_sizes = taosMemoryCalloc(num_keys * 2, sizeof(size_t));
+  char  **errs = taosMemoryCalloc(num_keys * 2, sizeof(char *));
+  rocksdb_multi_get(pTsdb->rCache.db, pTsdb->rCache.readoptions, num_keys * 2, (const char *const *)keys_list,
+                    keys_list_sizes, values_list, values_list_sizes, errs);
+  for (int i = 0; i < num_keys; ++i) {
+    taosMemoryFree(keys_list[i]);
+  }
+  taosMemoryFree(keys_list);
+  taosMemoryFree(keys_list_sizes);
+  taosMemoryFree(errs);
+
+  TSKEY                 keyTs = TSDBROW_TS(pRow);
+  rocksdb_writebatch_t *wb = pTsdb->rCache.writebatch;
+  for (int i = 0; i < num_keys; ++i) {
+    SColVal *pColVal = (SColVal *)taosArrayGet(aColVal, i);
+    if (COL_VAL_IS_VALUE(pColVal)) {
+      SLastCol *pLastCol = NULL;
+      if (NULL != values_list[i]) {
+        pLastCol = tsdbCacheDeserialize(values_list[i]);
+      }
+
+      if (NULL == pLastCol || pLastCol->ts <= keyTs) {
+        char  *value = NULL;
+        size_t vlen = 0;
+        tsdbCacheSerialize(&(SLastCol){.ts = keyTs, .colVal = *pColVal}, &value, &vlen);
+        char   key[ROCKS_KEY_LEN];
+        size_t klen = snprintf(key, ROCKS_KEY_LEN, "%" PRIi64 ":%" PRIi16 ":last", uid, pColVal->cid);
+        rocksdb_writebatch_put(wb, key, klen, value, vlen);
+        taosMemoryFree(value);
+      }
+    }
+
+    if (!COL_VAL_IS_NONE(pColVal)) {
+      SLastCol *pLastCol = NULL;
+      if (NULL != values_list[i + num_keys]) {
+        pLastCol = tsdbCacheDeserialize(values_list[i]);
+      }
+
+      if (NULL == pLastCol || pLastCol->ts <= keyTs) {
+        char  *value = NULL;
+        size_t vlen = 0;
+        tsdbCacheSerialize(&(SLastCol){.ts = keyTs, .colVal = *pColVal}, &value, &vlen);
+        char   key[ROCKS_KEY_LEN];
+        size_t klen = snprintf(key, ROCKS_KEY_LEN, "%" PRIi64 ":%" PRIi16 ":last_row", uid, pColVal->cid);
+        rocksdb_writebatch_put(wb, key, klen, value, vlen);
+        taosMemoryFree(value);
+      }
+    }
+  }
+  taosMemoryFree(values_list);
+  taosMemoryFree(values_list_sizes);
+
+  char *err = NULL;
+  rocksdb_write(pTsdb->rCache.db, pTsdb->rCache.writeoptions, wb, &err);
+  if (NULL != err) {
+    tsdbError("vgId:%d, %s failed at line %d since %s", TD_VID(pTsdb->pVnode), __func__, __LINE__, err);
+    rocksdb_free(err);
   }
-  rocksdb_put(pTsdb->rCache.db, pTsdb->rCache.writeoptions, "key", 3, "value", 5, &err);
+  rocksdb_writebatch_clear(wb);
 
+_exit:
+  taosArrayDestroy(aColVal);
+  taosMemoryFree(pTSchema);
   return code;
 }
 
diff --git a/source/dnode/vnode/src/tsdb/tsdbMemTable.c b/source/dnode/vnode/src/tsdb/tsdbMemTable.c
index f480fb6081d78305c3215f330fbcce03f87bf5bd..6b4ab4d5cf7cf5ae430f0eeb6e94d113bb6de141 100644
--- a/source/dnode/vnode/src/tsdb/tsdbMemTable.c
+++ b/source/dnode/vnode/src/tsdb/tsdbMemTable.c
@@ -678,7 +678,7 @@ static int32_t tsdbInsertColDataToTable(SMemTable *pMemTable, STbData *pTbData,
     tsdbCacheInsertLast(pMemTable->pTsdb->lruCache, pTbData->uid, &lRow, pMemTable->pTsdb);
   }
   */
-  tsdbCacheUpdate(pMemTable->pTsdb, pTbData->uid, &lRow);
+  tsdbCacheUpdate(pMemTable->pTsdb, pTbData->suid, pTbData->uid, &lRow);
 
   // SMemTable
   pMemTable->minKey = TMIN(pMemTable->minKey, pTbData->minKey);
@@ -748,7 +748,7 @@ static int32_t tsdbInsertRowDataToTable(SMemTable *pMemTable, STbData *pTbData,
     tsdbCacheInsertLast(pMemTable->pTsdb->lruCache, pTbData->uid, &lRow, pMemTable->pTsdb);
   }
   */
-  tsdbCacheUpdate(pMemTable->pTsdb, pTbData->uid, &lRow);
+  tsdbCacheUpdate(pMemTable->pTsdb, pTbData->suid, pTbData->uid, &lRow);
 
   // SMemTable
   pMemTable->minKey = TMIN(pMemTable->minKey, pTbData->minKey);