diff --git a/docs/en/solutions/How_to_Backup_and_Restore_ClickHouse_with_clickhouse_backup.md b/docs/en/solutions/How_to_Backup_and_Restore_ClickHouse_with_clickhouse_backup.md
new file mode 100644
index 00000000..dfc49e4d
--- /dev/null
+++ b/docs/en/solutions/How_to_Backup_and_Restore_ClickHouse_with_clickhouse_backup.md
@@ -0,0 +1,604 @@
+---
+kind:
+   - How To
+products:
+   - Alauda Container Platform
+   - Alauda Application Services
+ProductsVersion:
+   - 4.3.x
+---
+# How to Back Up and Restore ClickHouse with clickhouse-backup
+
+## Purpose
+
+This document explains how to back up and restore a ClickHouse instance deployed by the ClickHouse Operator on Alauda Container Platform by using a `clickhouse-backup` sidecar and S3-compatible object storage.
+
+The procedure covers:
+
+- Deploying ClickHouse with a `clickhouse-backup` sidecar.
+- Creating a remote backup in S3-compatible storage.
+- Running backups from a Kubernetes Job or CronJob.
+- Restoring the backup into a separate ClickHouseInstallation.
+- Verifying that restored tables contain the expected data.
+
+## Resolution
+
+### 1. Overview
+
+`clickhouse-backup` runs as a sidecar in the ClickHouse Pod. It connects to ClickHouse through `localhost:9000`, freezes MergeTree table parts, writes backup metadata, and uploads the backup to S3-compatible storage.
+
+For this workflow to back up table data correctly, the sidecar must mount the same ClickHouse data volume as the main `clickhouse` container at `/var/lib/clickhouse`. If the sidecar does not see the ClickHouse data directory, the backup may contain only metadata files.
+
+Distributed tables are backed up as schema only. The actual data is stored in the underlying MergeTree-family local tables, such as `events_local`.
+
+### 2. Prerequisites
+
+Prepare the following items before starting:
+
+| Item | Description | Example |
+|------|-------------|---------|
+| Namespace | Namespace for the ClickHouse instances and Jobs | `<namespace>` |
+| Source ClickHouseInstallation name | Source instance name | `<source-clickhouseinstallation-name>` |
+| Restore ClickHouseInstallation name | Restore instance name | `<restore-clickhouseinstallation-name>` |
+| ClickHouse cluster name | Cluster name in the ClickHouseInstallation spec | `<clickhouse-cluster-name>` |
+| Source ClickHouse Pod name | Source Pod generated by the ClickHouse Operator | `<source-clickhouse-pod-name>` |
+| Source ClickHouse service name | Service for the source Pod or shard | `<source-clickhouse-service-name>` |
+| Restore ClickHouse Pod name | Restore Pod generated by the ClickHouse Operator | `<restore-clickhouse-pod-name>` |
+| Restore ClickHouse service name | Service for the restore Pod or shard | `<restore-clickhouse-service-name>` |
+| S3 endpoint | S3-compatible object storage endpoint | `http://<s3-host>:<s3-port>` |
+| S3 bucket | Bucket for backup storage | `<s3-bucket>` |
+| S3 access key | Access key for the bucket | `<s3-access-key>` |
+| S3 secret key | Secret key for the bucket | `<s3-secret-key>` |
+| S3 credential Secret name | Kubernetes Secret that stores S3 credentials | `<s3-credential-secret-name>` |
+| Backup Job name | Kubernetes Job used to create a backup | `<backup-job-name>` |
+| Backup CronJob name | Kubernetes CronJob used for scheduled backups | `<backup-cronjob-name>` |
+| Restore Job name | Kubernetes Job used to restore a backup | `<restore-job-name>` |
+
+Set local variables for the commands in this document. If these variables are already exported in your shell, the manifest templates below can be rendered directly with `envsubst`.
+
+Use explicit variable lists with `envsubst`. This prevents runtime variables inside Job scripts, such as `$BACKUP_NAME`, `$COMMAND`, and `$STATUS`, from being replaced on your workstation.
+
+```bash
+export NAMESPACE="<namespace>"
+export SOURCE_CHI="<source-clickhouseinstallation-name>"
+export RESTORE_CHI="<restore-clickhouseinstallation-name>"
+export CLUSTER_NAME="<clickhouse-cluster-name>"
+export SOURCE_POD="<source-clickhouse-pod-name>"
+export SOURCE_SERVICE="<source-clickhouse-service-name>"
+export RESTORE_POD="<restore-clickhouse-pod-name>"
+export RESTORE_SERVICE="<restore-clickhouse-service-name>"
+export S3_ENDPOINT="http://<s3-host>:<s3-port>"
+export S3_BUCKET="<s3-bucket>"
+export S3_ACCESS_KEY="<s3-access-key>"
+export S3_SECRET_KEY="<s3-secret-key>"
+export S3_SECRET_NAME="<s3-credential-secret-name>"
+export BACKUP_JOB_NAME="<backup-job-name>"
+export BACKUP_CRONJOB_NAME="<backup-cronjob-name>"
+export RESTORE_JOB_NAME="<restore-job-name>"
+```
+
+Create the S3 bucket before running the backup workflow.
+
+```bash
+mc alias set backup-s3 "$S3_ENDPOINT" "$S3_ACCESS_KEY" "$S3_SECRET_KEY"
+mc mb --ignore-existing "backup-s3/$S3_BUCKET"
+```
+
+### 3. Create a Namespace and S3 Credential Secret
+
+```bash
+kubectl create namespace "$NAMESPACE"
+
+kubectl -n "$NAMESPACE" create secret generic "$S3_SECRET_NAME" \
+  --from-literal=access-key="$S3_ACCESS_KEY" \
+  --from-literal=secret-key="$S3_SECRET_KEY"
+```
+
+### 4. Deploy ClickHouse with a Backup Sidecar
+
+Create `clickhouse-source.yaml.tmpl`. The template uses the environment variables defined in the prerequisites section and can be rendered directly with `envsubst`.
+
+```yaml
+apiVersion: clickhouse.altinity.com/v1
+kind: ClickHouseInstallation
+metadata:
+  name: ${SOURCE_CHI}
+  namespace: ${NAMESPACE}
+spec:
+  configuration:
+    clusters:
+      - name: ${CLUSTER_NAME}
+        layout:
+          shardsCount: 1
+          replicasCount: 1
+        templates:
+          podTemplate: clickhouse-with-backup
+  templates:
+    podTemplates:
+      - name: clickhouse-with-backup
+        spec:
+          containers:
+            - name: clickhouse
+              volumeMounts:
+                - name: clickhouse-data
+                  mountPath: /var/lib/clickhouse
+            - name: clickhouse-backup
+              image: docker-mirrors.alauda.cn/altinity/clickhouse-backup:2.6.3
+              imagePullPolicy: IfNotPresent
+              args:
+                - server
+              env:
+                - name: LOG_LEVEL
+                  value: debug
+                - name: ALLOW_EMPTY_BACKUPS
+                  value: "false"
+                - name: API_LISTEN
+                  value: 0.0.0.0:7171
+                - name: API_CREATE_INTEGRATION_TABLES
+                  value: "true"
+                - name: BACKUPS_TO_KEEP_REMOTE
+                  value: "3"
+                - name: REMOTE_STORAGE
+                  value: s3
+                - name: S3_ACL
+                  value: private
+                - name: S3_ENDPOINT
+                  value: ${S3_ENDPOINT}
+                - name: S3_BUCKET
+                  value: ${S3_BUCKET}
+                - name: S3_PATH
+                  value: backup/shard-{shard}
+                - name: S3_ACCESS_KEY
+                  valueFrom:
+                    secretKeyRef:
+                      name: ${S3_SECRET_NAME}
+                      key: access-key
+                - name: S3_SECRET_KEY
+                  valueFrom:
+                    secretKeyRef:
+                      name: ${S3_SECRET_NAME}
+                      key: secret-key
+                - name: S3_FORCE_PATH_STYLE
+                  value: "true"
+                - name: S3_DISABLE_SSL
+                  value: "true"
+              ports:
+                - containerPort: 7171
+                  name: backup-rest
+              volumeMounts:
+                - name: clickhouse-data
+                  mountPath: /var/lib/clickhouse
+              securityContext:
+                runAsUser: 101
+                runAsGroup: 101
+    volumeClaimTemplates:
+      - name: clickhouse-data
+        spec:
+          accessModes:
+            - ReadWriteOnce
+          resources:
+            requests:
+              storage: 10Gi
+```
+
+Notes:
+
+- `ALLOW_EMPTY_BACKUPS="false"` is recommended for production because it fails when no data is present. Set it to `"true"` only for initial setup, CI, testing, or reusable templates where an empty database is expected.
+- `S3_DISABLE_SSL="true"` disables TLS for S3 traffic and sends backups without transport encryption. Use it only for local testing or trusted isolated networks. For production or untrusted networks, set it to `"false"` or omit it and configure TLS for the S3 endpoint.
+- Keep the `clickhouse` container entry in the Pod template. If it is omitted, the generated Pod may contain only the sidecar container.
+
+Apply the manifest and wait for the Pod to become ready.
+
+```bash
+envsubst '${SOURCE_CHI} ${NAMESPACE} ${CLUSTER_NAME} ${S3_ENDPOINT} ${S3_BUCKET} ${S3_SECRET_NAME}' \
+  < clickhouse-source.yaml.tmpl > clickhouse-source.yaml
+
+kubectl apply -f clickhouse-source.yaml
+kubectl -n "$NAMESPACE" wait --for=condition=Ready pod/"$SOURCE_POD" --timeout=10m
+kubectl -n "$NAMESPACE" get pod "$SOURCE_POD"
+```
+
+The Pod must show two ready containers.
+
+```text
+NAME                           READY   STATUS
+<source-clickhouse-pod-name>   2/2     Running
+```
+
+Verify that the backup integration tables were created.
+
+```bash
+kubectl -n "$NAMESPACE" exec "$SOURCE_POD" -c clickhouse -- \
+  clickhouse-client -q "SHOW TABLES FROM system LIKE 'backup_%'"
+```
+
+Expected output includes:
+
+```text
+backup_actions
+backup_list
+backup_version
+```
+
+### 5. Insert Test Data
+
+Create a local MergeTree table and a Distributed table.
+
+```bash
+kubectl -n "$NAMESPACE" exec "$SOURCE_POD" -c clickhouse -- \
+  clickhouse-client -mn --query "
+    CREATE TABLE events_local
+    (
+      event_date Date,
+      event_type Int32,
+      article_id Int32,
+      title String
+    )
+    ENGINE = MergeTree()
+    PARTITION BY toYYYYMM(event_date)
+    ORDER BY (event_type, article_id);
+
+    CREATE TABLE events AS events_local
+    ENGINE = Distributed('$CLUSTER_NAME', default, events_local, rand());
+
+    INSERT INTO events_local
+    SELECT today(), rand() % 3, number, 'backup test' FROM numbers(1000);
+
+    SELECT count() FROM events_local;
+  "
+```
+
+The expected count is:
+
+```text
+1000
+```
+
+Confirm that ClickHouse created active data parts for the local table.
+
+```bash
+kubectl -n "$NAMESPACE" exec "$SOURCE_POD" -c clickhouse -- \
+  clickhouse-client -q "
+    SELECT table, sum(rows), sum(bytes_on_disk)
+    FROM system.parts
+    WHERE database = 'default' AND table = 'events_local' AND active
+    GROUP BY table
+  "
+```
+
+### 6. Create and Upload a Backup Manually
+
+The `create_remote` command creates a local backup and uploads it to the configured remote storage in one step.
+
+```bash
+BACKUP_NAME="full-$(date +%Y%m%d%H%M%S)"
+
+kubectl -n "$NAMESPACE" exec "$SOURCE_POD" -c clickhouse-backup -- \
+  clickhouse-backup create_remote "$BACKUP_NAME"
+```
+
+List the backup from the sidecar.
+
+```bash
+kubectl -n "$NAMESPACE" exec "$SOURCE_POD" -c clickhouse-backup -- \
+  clickhouse-backup list | grep "$BACKUP_NAME"
+```
+
+Expected output contains both local and remote entries.
+
+```text
+<backup-name>   ...   local    regular
+<backup-name>   ...   remote   tar, regular
+```
+
+Verify that the remote storage contains both metadata and data part objects.
+
+```bash
+mc find "backup-s3/$S3_BUCKET/backup/shard-0/$BACKUP_NAME" --maxdepth 5
+```
+
+Expected objects include a `shadow` path for MergeTree data parts.
+
+```text
+backup/shard-0/<backup-name>/metadata.json
+backup/shard-0/<backup-name>/metadata/default/events.json
+backup/shard-0/<backup-name>/metadata/default/events_local.json
+backup/shard-0/<backup-name>/shadow/default/events_local/default_202605_1_1_0.tar
+```
+
+### 7. Run Backup from a Kubernetes Job
+
+The sidecar exposes `system.backup_actions`. This allows backup automation from a Kubernetes Job that only needs `clickhouse-client` network access to the ClickHouse service.
+
+Create `clickhouse-backup-job.yaml.tmpl`. Render only the manifest variables with `envsubst`; do not render runtime variables such as `$BACKUP_NAME`, `$COMMAND`, or `$STATUS` inside the Job script.
+
+```yaml
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: ${BACKUP_JOB_NAME}
+  namespace: ${NAMESPACE}
+spec:
+  backoffLimit: 1
+  template:
+    spec:
+      restartPolicy: Never
+      containers:
+        - name: run-backup
+          image: clickhouse/clickhouse-client:latest
+          imagePullPolicy: IfNotPresent
+          env:
+            - name: CLICKHOUSE_HOST
+              value: ${SOURCE_SERVICE}
+            - name: CLICKHOUSE_PORT
+              value: "9000"
+          command:
+            - bash
+            - -ec
+            - |
+              BACKUP_NAME="full-$(date +%Y%m%d%H%M%S)"
+              COMMAND="create_remote ${BACKUP_NAME}"
+
+              clickhouse-client --host="$CLICKHOUSE_HOST" --port="$CLICKHOUSE_PORT" -mn \
+                --query="INSERT INTO system.backup_actions(command) VALUES('${COMMAND}')"
+
+              while true; do
+                STATUS=$(clickhouse-client --host="$CLICKHOUSE_HOST" --port="$CLICKHOUSE_PORT" -mn \
+                  --query="SELECT status FROM system.backup_actions WHERE command='${COMMAND}' ORDER BY start DESC LIMIT 1 FORMAT TabSeparatedRaw")
+                echo "${COMMAND}: ${STATUS}"
+                if [ "$STATUS" != "in progress" ]; then
+                  break
+                fi
+                sleep 2
+              done
+
+              if [ "$STATUS" != "success" ]; then
+                clickhouse-client --host="$CLICKHOUSE_HOST" --port="$CLICKHOUSE_PORT" -mn \
+                  --query="SELECT command,status,error FROM system.backup_actions WHERE command='${COMMAND}' ORDER BY start DESC LIMIT 1"
+                exit 1
+              fi
+
+              echo "BACKUP_NAME=${BACKUP_NAME}"
+```
+
+Apply and verify the Job.
+
+```bash
+envsubst '${BACKUP_JOB_NAME} ${NAMESPACE} ${SOURCE_SERVICE}' \
+  < clickhouse-backup-job.yaml.tmpl > clickhouse-backup-job.yaml
+
+kubectl apply -f clickhouse-backup-job.yaml
+kubectl -n "$NAMESPACE" wait --for=condition=complete job/"$BACKUP_JOB_NAME" --timeout=20m
+kubectl -n "$NAMESPACE" logs job/"$BACKUP_JOB_NAME"
+```
+
+### 8. Schedule Backups with a CronJob
+
+After the manual Job succeeds, use the same command body in a CronJob.
+
+Create `clickhouse-backup-cronjob.yaml.tmpl` and render only the manifest variables with `envsubst`.
+
+```yaml
+apiVersion: batch/v1
+kind: CronJob
+metadata:
+  name: ${BACKUP_CRONJOB_NAME}
+  namespace: ${NAMESPACE}
+spec:
+  schedule: "0 0 * * *"
+  concurrencyPolicy: Forbid
+  successfulJobsHistoryLimit: 3
+  failedJobsHistoryLimit: 3
+  jobTemplate:
+    spec:
+      backoffLimit: 1
+      template:
+        spec:
+          restartPolicy: Never
+          containers:
+            - name: run-backup
+              image: clickhouse/clickhouse-client:latest
+              imagePullPolicy: IfNotPresent
+              env:
+                - name: CLICKHOUSE_HOST
+                  value: ${SOURCE_SERVICE}
+                - name: CLICKHOUSE_PORT
+                  value: "9000"
+              command:
+                - bash
+                - -ec
+                - |
+                  BACKUP_NAME="full-$(date +%Y%m%d%H%M%S)"
+                  COMMAND="create_remote ${BACKUP_NAME}"
+                  clickhouse-client --host="$CLICKHOUSE_HOST" --port="$CLICKHOUSE_PORT" -mn \
+                    --query="INSERT INTO system.backup_actions(command) VALUES('${COMMAND}')"
+                  while true; do
+                    STATUS=$(clickhouse-client --host="$CLICKHOUSE_HOST" --port="$CLICKHOUSE_PORT" -mn \
+                      --query="SELECT status FROM system.backup_actions WHERE command='${COMMAND}' ORDER BY start DESC LIMIT 1 FORMAT TabSeparatedRaw")
+                    echo "${COMMAND}: ${STATUS}"
+                    if [ "$STATUS" != "in progress" ]; then
+                      break
+                    fi
+                    sleep 2
+                  done
+                  test "$STATUS" = "success"
+```
+
+Render and apply the CronJob.
+
+```bash
+envsubst '${BACKUP_CRONJOB_NAME} ${NAMESPACE} ${SOURCE_SERVICE}' \
+  < clickhouse-backup-cronjob.yaml.tmpl > clickhouse-backup-cronjob.yaml
+
+kubectl apply -f clickhouse-backup-cronjob.yaml
+```
+
+For multi-shard clusters, run the same command once against one replica service per shard, and use a unique `S3_PATH` or backup name per shard.
+
+### 9. Create a Restore ClickHouse Instance
+
+Create another ClickHouseInstallation with the same sidecar and S3 configuration. Use a different instance name.
+
+Create `clickhouse-restore.yaml.tmpl` by copying `clickhouse-source.yaml.tmpl` and changing the metadata name variable from `${SOURCE_CHI}` to `${RESTORE_CHI}`. Then render the template.
+
+```bash
+envsubst '${RESTORE_CHI} ${NAMESPACE} ${CLUSTER_NAME} ${S3_ENDPOINT} ${S3_BUCKET} ${S3_SECRET_NAME}' \
+  < clickhouse-restore.yaml.tmpl > clickhouse-restore.yaml
+
+kubectl apply -f clickhouse-restore.yaml
+kubectl -n "$NAMESPACE" wait --for=condition=Ready pod/"$RESTORE_POD" --timeout=10m
+```
+
+### 10. Restore the Backup
+
+Use the backup name from step 6 or step 7.
+
+```bash
+BACKUP_NAME="<backup-name>"
+
+kubectl -n "$NAMESPACE" exec "$RESTORE_POD" -c clickhouse-backup -- \
+  clickhouse-backup restore_remote "$BACKUP_NAME"
+```
+
+The same restore can be automated through `system.backup_actions`.
+
+Create `clickhouse-restore-job.yaml.tmpl`. Set `BACKUP_NAME` to the backup created in step 6 or step 7, and render only the manifest variables with `envsubst`.
+
+```yaml
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: ${RESTORE_JOB_NAME}
+  namespace: ${NAMESPACE}
+spec:
+  backoffLimit: 0
+  template:
+    spec:
+      restartPolicy: Never
+      containers:
+        - name: run-restore
+          image: clickhouse/clickhouse-client:latest
+          imagePullPolicy: IfNotPresent
+          env:
+            - name: CLICKHOUSE_HOST
+              value: ${RESTORE_SERVICE}
+            - name: CLICKHOUSE_PORT
+              value: "9000"
+            - name: RESTORE_BACKUP
+              value: ${BACKUP_NAME}
+          command:
+            - bash
+            - -ec
+            - |
+              COMMAND="restore_remote ${RESTORE_BACKUP}"
+              clickhouse-client --host="$CLICKHOUSE_HOST" --port="$CLICKHOUSE_PORT" -mn \
+                --query="INSERT INTO system.backup_actions(command) VALUES('${COMMAND}')"
+              while true; do
+                STATUS=$(clickhouse-client --host="$CLICKHOUSE_HOST" --port="$CLICKHOUSE_PORT" -mn \
+                  --query="SELECT status FROM system.backup_actions WHERE command='${COMMAND}' ORDER BY start DESC LIMIT 1 FORMAT TabSeparatedRaw")
+                echo "${COMMAND}: ${STATUS}"
+                if [ "$STATUS" != "in progress" ]; then
+                  break
+                fi
+                sleep 2
+              done
+              test "$STATUS" = "success"
+```
+
+Apply and verify the restore Job.
+
+```bash
+envsubst '${RESTORE_JOB_NAME} ${NAMESPACE} ${RESTORE_SERVICE} ${BACKUP_NAME}' \
+  < clickhouse-restore-job.yaml.tmpl > clickhouse-restore-job.yaml
+
+kubectl apply -f clickhouse-restore-job.yaml
+kubectl -n "$NAMESPACE" wait --for=condition=complete job/"$RESTORE_JOB_NAME" --timeout=20m
+kubectl -n "$NAMESPACE" logs job/"$RESTORE_JOB_NAME"
+```
+
+### 11. Verify Restored Data
+
+```bash
+kubectl -n "$NAMESPACE" exec "$RESTORE_POD" -c clickhouse -- \
+  clickhouse-client -mn --query "
+    SHOW TABLES;
+    SELECT count() FROM events_local;
+    SELECT count() FROM events;
+  "
+```
+
+Expected output:
+
+```text
+events
+events_local
+1000
+1000
+```
+
+### 12. Troubleshooting
+
+#### The Pod shows only one container
+
+If the Pod shows `1/1` and only the `clickhouse-backup` container exists, the custom Pod template replaced the default ClickHouse container. Add the `- name: clickhouse` container entry to the Pod template.
+
+```bash
+kubectl -n "$NAMESPACE" get pod "$SOURCE_POD"
+```
+
+#### The backup contains only metadata files
+
+Check whether both containers mount the same volume at `/var/lib/clickhouse`.
+
+```bash
+kubectl -n "$NAMESPACE" get pod "$SOURCE_POD" -o jsonpath='{range .spec.containers[*]}{.name}{"\n"}{range .volumeMounts[*]}{.name}{" -> "}{.mountPath}{"\n"}{end}{end}'
+```
+
+A valid data backup contains `shadow/...tar` objects in S3. A backup that only contains `metadata/*.json` objects is schema-only or cannot see the ClickHouse data directory.
+
+```bash
+mc find "backup-s3/$S3_BUCKET/backup/shard-0/<backup-name>" --maxdepth 5
+```
+
+#### The backup integration tables do not exist
+
+Check the `clickhouse-backup` container logs.
+
+```bash
+kubectl -n "$NAMESPACE" logs "$SOURCE_POD" -c clickhouse-backup --tail=100
+```
+
+Also confirm that `API_CREATE_INTEGRATION_TABLES` is set to `true`.
+
+#### Distributed table data is not included
+
+This is expected. Distributed tables are backed up as schema only. Back up and restore the underlying MergeTree-family local tables.
+
+#### Restore cannot find the backup
+
+Confirm that the restore instance uses the same `S3_ENDPOINT`, `S3_BUCKET`, and `S3_PATH` as the source instance.
+
+```bash
+kubectl -n "$NAMESPACE" exec "$RESTORE_POD" -c clickhouse-backup -- \
+  clickhouse-backup list
+```
+
+### 13. Validation Result
+
+This procedure was validated on an Alauda Container Platform 4.3 environment with ClickHouse Operator 4.3 and `altinity/clickhouse-backup:2.6.3`.
+
+Validated results:
+
+- The source Pod ran with two containers: `clickhouse` and `clickhouse-backup`.
+- Both containers mounted the same `/var/lib/clickhouse` volume.
+- `system.backup_actions`, `system.backup_list`, and `system.backup_version` were created.
+- `clickhouse-backup create_remote` uploaded metadata and MergeTree data part archives to S3.
+- `system.backup_actions` successfully triggered `create_remote <backup-name>`.
+- `clickhouse-backup restore_remote <backup-name>` restored the backup into a separate ClickHouseInstallation.
+- The restored `events_local` and `events` tables both returned `1000` rows.
+
+## Related Information
+
+- `clickhouse-backup` supports backup and restore for MergeTree-family table engines.
+- Use a separate restore ClickHouseInstallation when validating a backup to avoid modifying the source instance.
+- For multi-shard clusters, run backup and restore operations once per shard and avoid restoring the same shard data on multiple replicas.