diff --git a/be/src/common/config.h b/be/src/common/config.h
index 59ab5e4020ec5f..cccb49ed002be3 100644
--- a/be/src/common/config.h
+++ b/be/src/common/config.h
@@ -703,8 +703,8 @@ CONF_Int32(metric_late_materialization_ratio, "1000");
 
 // Max batched bytes for each transmit request. (256KB)
 CONF_Int64(max_transmit_batched_bytes, "262144");
-// max chunk size for each tablet write request. (128MB)
-CONF_mInt64(max_tablet_write_chunk_bytes, "134217728");
+// max chunk size for each tablet write request. (512MB)
+CONF_mInt64(max_tablet_write_chunk_bytes, "536870912");
 
 CONF_Int16(bitmap_max_filter_items, "30");
 
diff --git a/be/src/exec/tablet_sink_index_channel.cpp b/be/src/exec/tablet_sink_index_channel.cpp
index 277bf3640388d1..76f27b9ae2b6f9 100644
--- a/be/src/exec/tablet_sink_index_channel.cpp
+++ b/be/src/exec/tablet_sink_index_channel.cpp
@@ -467,8 +467,8 @@ Status NodeChannel::add_chunk(Chunk* input, const std::vector<int64_t>& tablet_i
         }
     }
 
-    if (_cur_chunk->num_rows() < _runtime_state->chunk_size() &&
-        _cur_chunk_mem_usage < config::max_tablet_write_chunk_bytes) {
+    if (_cur_chunk->num_rows() <=0 || (_cur_chunk->num_rows() < _runtime_state->chunk_size() &&
+        _cur_chunk_mem_usage < config::max_tablet_write_chunk_bytes)) {
         // 2. chunk not full
         if (_request_queue.empty()) {
             return Status::OK();
@@ -520,8 +520,8 @@ Status NodeChannel::add_chunks(Chunk* input, const std::vector<std::vector<int64
         }
     }
 
-    if (_cur_chunk->num_rows() < _runtime_state->chunk_size() &&
-        _cur_chunk_mem_usage < config::max_tablet_write_chunk_bytes) {
+    if (_cur_chunk->num_rows() <= 0 || (_cur_chunk->num_rows() < _runtime_state->chunk_size() &&
+        _cur_chunk_mem_usage < config::max_tablet_write_chunk_bytes)) {
         // 2. chunk not full
         if (_request_queue.empty()) {
             return Status::OK();