diff --git a/deployments/ticdc/docker-compose/docker-compose-kafka-integration.yml b/deployments/ticdc/docker-compose/docker-compose-kafka-integration.yml
index 2e794be09db..d237bf12bcf 100644
--- a/deployments/ticdc/docker-compose/docker-compose-kafka-integration.yml
+++ b/deployments/ticdc/docker-compose/docker-compose-kafka-integration.yml
@@ -11,8 +11,9 @@ services:
     ports:
       - "9092:9092"
     environment:
-      KAFKA_MESSAGE_MAX_BYTES: 1073741824
-      KAFKA_REPLICA_FETCH_MAX_BYTES: 1073741824
+      KAFKA_MESSAGE_MAX_BYTES: 11534336
+      KAFKA_REPLICA_FETCH_MAX_BYTES: 11534336
+      KAFKA_CREATE_TOPICS: "big-message-test:1:1"
       KAFKA_BROKER_ID: 1
       RACK_COMMAND: "curl -sfL https://git.io/JJZXX -o /tmp/kafka.server.keystore.jks && curl -sfL https://git.io/JJZXM -o /tmp/kafka.server.truststore.jks"
       KAFKA_LISTENERS: "SSL://127.0.0.1:9093,PLAINTEXT://127.0.0.1:9092"
diff --git a/tests/integration_tests/kafka_big_messages/conf/diff_config.toml b/tests/integration_tests/kafka_big_messages/conf/diff_config.toml
new file mode 100644
index 00000000000..0082a370281
--- /dev/null
+++ b/tests/integration_tests/kafka_big_messages/conf/diff_config.toml
@@ -0,0 +1,29 @@
+# diff Configuration.
+
+check-thread-count = 4
+
+export-fix-sql = true
+
+check-struct-only = false
+
+[task]
+output-dir = "/tmp/tidb_cdc_test/kafka_big_messages/sync_diff/output"
+
+source-instances = ["mysql1"]
+
+target-instance = "tidb0"
+
+target-check-tables = ["kafka_big_messages.test"]
+
+[data-sources]
+[data-sources.mysql1]
+host = "127.0.0.1"
+port = 4000
+user = "root"
+password = ""
+
+[data-sources.tidb0]
+host = "127.0.0.1"
+port = 3306
+user = "root"
+password = ""
diff --git a/tests/integration_tests/kafka_big_messages/run.sh b/tests/integration_tests/kafka_big_messages/run.sh
new file mode 100755
index 00000000000..0b10c780c43
--- /dev/null
+++ b/tests/integration_tests/kafka_big_messages/run.sh
@@ -0,0 +1,55 @@
+#!/bin/bash
+
+set -e
+
+CUR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+source $CUR/../_utils/test_prepare
+WORK_DIR=$OUT_DIR/$TEST_NAME
+CDC_BINARY=cdc.test
+SINK_TYPE=$1
+
+function run() {
+	# test kafka sink only in this case
+	if [ "$SINK_TYPE" == "mysql" ]; then
+		return
+	fi
+	rm -rf $WORK_DIR && mkdir -p $WORK_DIR
+
+	start_tidb_cluster --workdir $WORK_DIR
+
+	cd $WORK_DIR
+
+	# record tso before we create tables to skip the system table DDLs
+	start_ts=$(cdc cli tso query --pd=http://$UP_PD_HOST_1:$UP_PD_PORT_1)
+
+	run_cdc_server --workdir $WORK_DIR --binary $CDC_BINARY
+
+	# Use a max-message-bytes parameter that is larger than the kafka topic max message bytes.
+	# Test if TiCDC automatically uses the max-message-bytes of the topic.
+	# See: https://github.com/PingCAP-QE/ci/blob/ddde195ebf4364a0028d53405d1194aa37a4d853/jenkins/pipelines/ci/ticdc/cdc_ghpr_kafka_integration_test.groovy#L178
+	# Use a topic that has already been created.
+	# See: https://github.com/PingCAP-QE/ci/blob/ddde195ebf4364a0028d53405d1194aa37a4d853/jenkins/pipelines/ci/ticdc/cdc_ghpr_kafka_integration_test.groovy#L180
+	SINK_URI="kafka://127.0.0.1:9092/big-message-test?protocol=open-protocol&partition-num=1&kafka-version=${KAFKA_VERSION}&max-message-bytes=12582912"
+	cdc cli changefeed create --start-ts=$start_ts --sink-uri="$SINK_URI"
+	run_kafka_consumer $WORK_DIR "kafka://127.0.0.1:9092/big-message-test?protocol=open-protocol&partition-num=1&version=${KAFKA_VERSION}"
+
+	echo "Starting generate kafka big messages..."
+	cd $CUR/../../utils/gen_kafka_big_messages
+	if [ ! -f ./gen_kafka_big_messages ]; then
+		GO111MODULE=on go build
+	fi
+	# Generate data larger than kafka broker max.message.bytes. We can send this data correctly.
+	./gen_kafka_big_messages --row-count=15 --sql-file-path=$CUR/test.sql
+
+	run_sql_file $CUR/test.sql ${UP_TIDB_HOST} ${UP_TIDB_PORT}
+	table="kafka_big_messages.test"
+	check_table_exists $table ${DOWN_TIDB_HOST} ${DOWN_TIDB_PORT}
+	check_sync_diff $WORK_DIR $CUR/conf/diff_config.toml
+
+	cleanup_process $CDC_BINARY
+}
+
+trap stop_tidb_cluster EXIT
+run $*
+check_logs $WORK_DIR
+echo "[$(date)] <<<<<< run test case $TEST_NAME success! >>>>>>"
diff --git a/tests/integration_tests/kafka_messages/conf/diff_config.toml b/tests/integration_tests/kafka_messages/conf/diff_config.toml
index f4a6d29c149..f471166e80c 100644
--- a/tests/integration_tests/kafka_messages/conf/diff_config.toml
+++ b/tests/integration_tests/kafka_messages/conf/diff_config.toml
@@ -7,7 +7,7 @@ export-fix-sql = true
 check-struct-only = false
 
 [task]
-    output-dir = "/tmp/tidb_cdc_test/kafka_message/sync_diff/output"
+    output-dir = "/tmp/tidb_cdc_test/kafka_messages/sync_diff/output"
 
     source-instances = ["mysql1"]
 
diff --git a/tests/integration_tests/kafka_messages/run.sh b/tests/integration_tests/kafka_messages/run.sh
index 1fb75b5adfa..5f89c04b421 100755
--- a/tests/integration_tests/kafka_messages/run.sh
+++ b/tests/integration_tests/kafka_messages/run.sh
@@ -31,10 +31,13 @@ function run_length_limit() {
 	run_cdc_server --workdir $WORK_DIR --binary $CDC_BINARY --loglevel "info"
 
 	TOPIC_NAME="ticdc-kafka-message-test-$RANDOM"
-	SINK_URI="kafka://127.0.0.1:9092/$TOPIC_NAME?protocol=open-protocol&partition-num=4&kafka-version=${KAFKA_VERSION}&max-message-bytes=10485760"
+	# Use a max-message-bytes parameter that is larger than the kafka broker max message bytes.
+	# Test if TiCDC automatically uses the max-message-bytes of the broker.
+	# See: https://github.com/PingCAP-QE/ci/blob/ddde195ebf4364a0028d53405d1194aa37a4d853/jenkins/pipelines/ci/ticdc/cdc_ghpr_kafka_integration_test.groovy#L178
+	SINK_URI="kafka://127.0.0.1:9092/$TOPIC_NAME?protocol=open-protocol&partition-num=4&kafka-version=${KAFKA_VERSION}&max-message-bytes=12582912"
 	run_cdc_cli changefeed create --start-ts=$start_ts --sink-uri="$SINK_URI"
 	if [ "$SINK_TYPE" == "kafka" ]; then
-		run_kafka_consumer $WORK_DIR "kafka://127.0.0.1:9092/$TOPIC_NAME?protocol=open-protocol&partition-num=4&version=${KAFKA_VERSION}&max-message-bytes=10485760"
+		run_kafka_consumer $WORK_DIR "kafka://127.0.0.1:9092/$TOPIC_NAME?protocol=open-protocol&partition-num=4&version=${KAFKA_VERSION}"
 	fi
 
 	# Add a check table to reduce check time, or if we check data with sync diff
diff --git a/tests/utils/gen_kafka_big_messages/main.go b/tests/utils/gen_kafka_big_messages/main.go
new file mode 100644
index 00000000000..f3675f44743
--- /dev/null
+++ b/tests/utils/gen_kafka_big_messages/main.go
@@ -0,0 +1,145 @@
+// Copyright 2021 PingCAP, Inc.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package main
+
+import (
+	"flag"
+	"fmt"
+	"log"
+	"os"
+	"strings"
+
+	"github.com/pingcap/errors"
+)
+
+// See: https://docs.pingcap.com/tidb/stable/tidb-limitations/#limitations-on-string-types
+const varcharColumnMaxLen = 16383
+
+// Value of col. Defined as a variable for testing.
+var colValue = strings.Repeat("a", varcharColumnMaxLen)
+
+type options struct {
+	// The size of each row.
+	// The default is 1MiB.
+	// FIXME: Currently it does not have precise control over the size of each row.
+	// The overhead needs to be calculated and processed accurately.
+	rowBytes int
+	// Total number of rows.
+	// The default is 1 line.
+	rowCount int
+	// Sql file path.
+	// The default is `./test.sql`.
+	sqlFilePath string
+	// Database name.
+	// The default is `kafka_big_messages`.
+	databaseName string
+	// Table name.
+	// The default is `test`.
+	tableName string
+}
+
+func (o *options) validate() error {
+	if o.rowBytes <= 0 {
+		return errors.New("rowBytes must be greater than zero")
+	}
+
+	if o.rowCount <= 0 {
+		return errors.New("rowCount must be greater than zero")
+	}
+
+	if o.sqlFilePath == "" {
+		return errors.New("please specify the correct file path")
+	}
+
+	if o.databaseName == "" {
+		return errors.New("please specify the database name")
+	}
+
+	if o.tableName == "" {
+		return errors.New("please specify the table name")
+	}
+
+	return nil
+}
+
+func gatherOptions() *options {
+	o := &options{}
+
+	fs := flag.NewFlagSet(os.Args[0], flag.ExitOnError)
+	fs.IntVar(&o.rowBytes, "row-bytes", 1024*1024, "Number of bytes per row.")
+	fs.IntVar(&o.rowCount, "row-count", 1, "Count of rows.")
+	fs.StringVar(&o.sqlFilePath, "sql-file-path", "./test.sql", "Sql file path.")
+	fs.StringVar(&o.databaseName, "database-name", "kafka_big_messages", "Database name.")
+	fs.StringVar(&o.tableName, "table-name", "test", "Table name.")
+
+	_ = fs.Parse(os.Args[1:])
+	return o
+}
+
+func main() {
+	o := gatherOptions()
+	if err := o.validate(); err != nil {
+		log.Panicf("Invalid options: %v", err)
+	}
+
+	file, err := os.OpenFile(o.sqlFilePath, os.O_CREATE|os.O_RDWR, os.ModePerm)
+	if err != nil {
+		log.Panicf("Open sql file failed: %v", err)
+	}
+
+	_, err = file.Write([]byte(genDatabaseSql(o.databaseName)))
+	if err != nil {
+		log.Panicf("Wirte create database sql failed: %v", err)
+	}
+
+	_, err = file.Write([]byte(genCreateTableSql(o.rowBytes, o.tableName)))
+	if err != nil {
+		log.Panicf("Wirte create table sql failed: %v", err)
+	}
+
+	for i := 0; i < o.rowCount; i++ {
+		_, err = file.Write([]byte(genInsertSql(o.rowBytes, o.tableName, i)))
+		if err != nil {
+			log.Panicf("Wirte insert sql failed: %v", err)
+		}
+	}
+}
+
+func genDatabaseSql(databaseName string) string {
+	return fmt.Sprintf(`DROP DATABASE IF EXISTS %s;
+CREATE DATABASE %s;
+USE %s;
+
+`, databaseName, databaseName, databaseName)
+}
+
+func genCreateTableSql(rawBytes int, tableName string) string {
+	var cols string
+
+	for i := 0; i < rawBytes/varcharColumnMaxLen; i++ {
+		cols = fmt.Sprintf("%s, a%d VARCHAR(%d)", cols, i, varcharColumnMaxLen)
+	}
+
+	return fmt.Sprintf("CREATE TABLE %s(id int primary key %s);\n", tableName, cols)
+}
+
+func genInsertSql(rawBytes int, tableName string, id int) string {
+	var values string
+
+	for i := 0; i < rawBytes/varcharColumnMaxLen; i++ {
+		values = fmt.Sprintf("%s, '%s'", values, colValue)
+	}
+
+	return fmt.Sprintf("INSERT INTO %s VALUES (%d%s);\n", tableName, id, values)
+}
diff --git a/tests/utils/gen_kafka_big_messages/main_test.go b/tests/utils/gen_kafka_big_messages/main_test.go
new file mode 100644
index 00000000000..504829d8f7d
--- /dev/null
+++ b/tests/utils/gen_kafka_big_messages/main_test.go
@@ -0,0 +1,121 @@
+// Copyright 2021 PingCAP, Inc.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package main
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/require"
+)
+
+func TestValidateOptions(t *testing.T) {
+	testCases := []struct {
+		o           *options
+		expectedErr string
+	}{
+		{
+			&options{
+				rowBytes: 0,
+			},
+			".*rowBytes must be greater than zero.*",
+		},
+		{
+			&options{
+				rowBytes: 1024 * 1024,
+				rowCount: 0,
+			},
+			".*rowCount must be greater than zero.*",
+		},
+		{
+			&options{
+				rowBytes:    1024 * 1024,
+				rowCount:    1,
+				sqlFilePath: "",
+			},
+			".*please specify the correct file path.*",
+		},
+		{
+			&options{
+				rowBytes:     1024 * 1024,
+				rowCount:     1,
+				sqlFilePath:  "./test.sql",
+				databaseName: "",
+			},
+			".*please specify the database name.*",
+		},
+		{
+			&options{
+				rowBytes:     1024 * 1024,
+				rowCount:     1,
+				sqlFilePath:  "./test.sql",
+				databaseName: "kafka-big-messages",
+				tableName:    "",
+			},
+			".*please specify the table name.*",
+		},
+		{
+			&options{
+				rowBytes:     1024 * 1024,
+				rowCount:     10,
+				sqlFilePath:  "./test.sql",
+				databaseName: "kafka-big-messages",
+				tableName:    "test",
+			},
+			"",
+		},
+	}
+
+	for _, tc := range testCases {
+		err := tc.o.validate()
+		if tc.expectedErr != "" {
+			require.Error(t, err)
+			require.Regexp(t, tc.expectedErr, tc.o.validate().Error())
+		} else {
+			require.Nil(t, err)
+		}
+	}
+}
+
+func TestGenDatabaseSql(t *testing.T) {
+	database := "test"
+
+	sql := genDatabaseSql(database)
+
+	require.Equal(t, "DROP DATABASE IF EXISTS test;\nCREATE DATABASE test;\nUSE test;\n\n", sql)
+}
+
+func TestGenCreateTableSql(t *testing.T) {
+	rawBytes := varcharColumnMaxLen
+	tableName := "test"
+
+	sql := genCreateTableSql(rawBytes, tableName)
+	require.Equal(t, "CREATE TABLE test(id int primary key , a0 VARCHAR(16383));\n", sql)
+}
+
+func TestGenInsertSql(t *testing.T) {
+	// Override the col value to test.
+	oldColValue := colValue
+	colValue = "a"
+	defer func() {
+		colValue = oldColValue
+	}()
+
+	rawBytes := varcharColumnMaxLen
+	tableName := "test"
+	id := 1
+
+	sql := genInsertSql(rawBytes, tableName, id)
+	println(sql)
+	require.Equal(t, "INSERT INTO test VALUES (1, 'a');\n", sql)
+}