diff --git a/api/v1alpha1/mysqlcluster_types.go b/api/v1alpha1/mysqlcluster_types.go
index f4414d66..ed4135d7 100644
--- a/api/v1alpha1/mysqlcluster_types.go
+++ b/api/v1alpha1/mysqlcluster_types.go
@@ -92,6 +92,11 @@ type MysqlClusterSpec struct {
 	// +optional
 	BackupSchedule string `json:"backupSchedule,omitempty"`
 
+	// Specify that crontab job backup both on NFS and S3 storage.
+	// +optional
+	// +kubebuilder:default:=false
+	BothS3NFS bool `json:"bothS3NFS,omitempty"`
+
 	// If set keeps last BackupScheduleJobsHistoryLimit Backups
 	// +optional
 	// +kubebuilder:default:=6
diff --git a/api/v1alpha1/mysqlcluster_webhook.go b/api/v1alpha1/mysqlcluster_webhook.go
index 61234792..961eb2a4 100644
--- a/api/v1alpha1/mysqlcluster_webhook.go
+++ b/api/v1alpha1/mysqlcluster_webhook.go
@@ -54,6 +54,11 @@ func (r *MysqlCluster) ValidateCreate() error {
 	if err := r.validateNFSServerAddress(r); err != nil {
 		return err
 	}
+
+	if err := r.validBothS3NFS(); err != nil {
+		return err
+	}
+
 	if err := r.validateMysqlVersionAndImage(); err != nil {
 		return err
 	}
@@ -74,6 +79,11 @@ func (r *MysqlCluster) ValidateUpdate(old runtime.Object) error {
 	if err := r.validateLowTableCase(oldCluster); err != nil {
 		return err
 	}
+
+	if err := r.validBothS3NFS(); err != nil {
+		return err
+	}
+
 	if err := r.validateMysqlVersionAndImage(); err != nil {
 		return err
 	}
@@ -145,3 +155,13 @@ func (r *MysqlCluster) validateMysqlVersionAndImage() error {
 	}
 	return nil
 }
+
+// Validate BothS3NFS
+func (r *MysqlCluster) validBothS3NFS() error {
+	if r.Spec.BothS3NFS &&
+		(len(r.Spec.NFSServerAddress) == 0 || len(r.Spec.BackupSchedule) == 0 ||
+			len(r.Spec.BackupSecretName) == 0) {
+		return apierrors.NewForbidden(schema.GroupResource{}, "", fmt.Errorf("if BothS3NFS is set, backupSchedule/backupSecret/nfsAddress should not empty"))
+	}
+	return nil
+}
diff --git a/api/v1alpha1/zz_generated.deepcopy.go b/api/v1alpha1/zz_generated.deepcopy.go
index bbc85fe9..2bbde94f 100644
--- a/api/v1alpha1/zz_generated.deepcopy.go
+++ b/api/v1alpha1/zz_generated.deepcopy.go
@@ -1,4 +1,3 @@
-//go:build !ignore_autogenerated
 // +build !ignore_autogenerated
 
 /*
diff --git a/backup/cronbackup.go b/backup/cronbackup.go
index ca00b238..9ef05521 100644
--- a/backup/cronbackup.go
+++ b/backup/cronbackup.go
@@ -24,6 +24,7 @@ type CronJob struct {
 	BackupScheduleJobsHistoryLimit *int
 	Image                          string
 	NFSServerAddress               string
+	BothS3NFS                      bool
 	Log                            logr.Logger
 }
 
@@ -45,9 +46,9 @@ func (j *CronJob) Run() {
 		log.Info("at least a backup is running", "running_backups_count", j.scheduledBackupsRunningCount())
 		return
 	}
-
+	//TODO: if BothS3NFS, it need create backup without nfs address.
 	// create the backup
-	if _, err := j.createBackup(); err != nil {
+	if err := j.createBackups(); err != nil {
 		log.Error(err, "failed to create backup")
 	}
 }
@@ -117,8 +118,32 @@ func (j *CronJob) backupGC() {
 	}
 }
 
-func (j *CronJob) createBackup() (*apiv1alpha1.Backup, error) {
-	backupName := fmt.Sprintf("%s-auto-%s", j.ClusterName, time.Now().Format("2006-01-02t15-04-05"))
+func (j *CronJob) createBackups() error {
+	// Cannot use uppcase, because RFC 1123 subdomain must consist of lower case alphanumeric characters
+	const nfstype string = "nfs"
+	const s3type string = "s3"
+	if j.BothS3NFS { // create both.
+		if err := j.createOneBackup(nfstype, j.NFSServerAddress); err != nil {
+			return err
+		}
+		if err := j.createOneBackup(s3type, ""); err != nil {
+			return err
+		}
+	} else {
+		if len(j.NFSServerAddress) > 0 {
+			if err := j.createOneBackup(nfstype, j.NFSServerAddress); err != nil {
+				return err
+			}
+		} else {
+			if err := j.createOneBackup(s3type, ""); err != nil {
+				return err
+			}
+		}
+	}
+	return nil
+}
+func (j *CronJob) createOneBackup(BackupType, NFSAddress string) error {
+	backupName := fmt.Sprintf("%s-%s-%s", j.ClusterName, BackupType, time.Now().Format("2006-01-02t15-04-05"))
 
 	backup := &apiv1alpha1.Backup{
 		ObjectMeta: metav1.ObjectMeta{
@@ -131,13 +156,12 @@ func (j *CronJob) createBackup() (*apiv1alpha1.Backup, error) {
 			//TODO modify to cluster sidecar image
 			Image: j.Image,
 			//RemoteDeletePolicy: j.BackupRemoteDeletePolicy,
-			HostName: fmt.Sprintf("%s-mysql-0", j.ClusterName),
+			HostName:         fmt.Sprintf("%s-mysql-0", j.ClusterName),
+			NFSServerAddress: NFSAddress,
 		},
 	}
-	if len(j.NFSServerAddress) > 0 {
-		backup.Spec.NFSServerAddress = j.NFSServerAddress
-	}
-	return backup, j.Client.Create(context.TODO(), backup)
+
+	return j.Client.Create(context.TODO(), backup)
 }
 
 type byTimestamp []apiv1alpha1.Backup
diff --git a/config/crd/bases/mysql.radondb.com_mysqlclusters.yaml b/config/crd/bases/mysql.radondb.com_mysqlclusters.yaml
index 5d56ba83..8c3d3c1c 100644
--- a/config/crd/bases/mysql.radondb.com_mysqlclusters.yaml
+++ b/config/crd/bases/mysql.radondb.com_mysqlclusters.yaml
@@ -70,6 +70,10 @@ spec:
                 description: Represents the name of the secret that contains credentials
                   to connect to the storage provider to store backups.
                 type: string
+              bothS3NFS:
+                default: false
+                description: Specify that crontab job backup both on NFS and S3 storage.
+                type: boolean
               metricsOpts:
                 default:
                   enabled: false
diff --git a/controllers/backupcron_controller.go b/controllers/backupcron_controller.go
index ae315e7d..11d10532 100644
--- a/controllers/backupcron_controller.go
+++ b/controllers/backupcron_controller.go
@@ -144,7 +144,8 @@ func (r *BackupCronReconciler) updateClusterSchedule(ctx context.Context, cluste
 		Image:                          cluster.Spec.PodPolicy.SidecarImage,
 		BackupScheduleJobsHistoryLimit: cluster.Spec.BackupScheduleJobsHistoryLimit,
 		//BackupRemoteDeletePolicy:       cluster.Spec.BackupRemoteDeletePolicy,
-
+		// add both S3 and NFS flag.
+		BothS3NFS:        cluster.Spec.BothS3NFS,
 		NFSServerAddress: cluster.Spec.NFSServerAddress,
 		Log:              log,
 	}, cluster.Name)