Backup and recovery for SAP HANA by using disk snapshots

This planning guide describes the disk snapshot feature of Google Cloud's Agent for SAP. For information about the other features, see Google Cloud's Agent for SAP planning guide.

The disk snapshot feature lets you perform backup and recovery operations for SAP HANA scale-up systems running on Compute Engine VM instances.

This feature is integrated with the Data snapshots feature of SAP HANA, which lets you to use standard snapshots of your /hana/data disk to back up and recover your SAP HANA database.

For information about how to back up and recover your SAP HANA database by using this feature of the agent, see Back up and recover SAP HANA by using a disk snapshot.

Monthly cost estimate

You incur charges for the standard snapshots that you create. For pricing information, see Disk snapshots.

To estimate the monthly costs for using disk snapshots as backups for your SAP HANA data volume, you can use the Google Cloud Pricing Calculator. Use the following information to help you better estimate the cost:

• Estimated full size for the initial disk snapshot.
• Estimated delta size for each disk snapshot.
• Snapshot frequency. For example, if you create one snapshot every day, then the snapshot frequency is 24 hours.
• Snapshot retention time.
• Snapshot storage redundancy: single-region or multi-region.
• Estimated intra-region and inter-region data transfer over your network. For example, you may incur network fees if you create or restore a standard snapshot that is stored in a different location than the source disk.

Limitations

The following limitations apply to snapshot based backup and recovery when using Google Cloud's Agent for SAP:

• You cannot use the agent's disk snapshot feature for SAP HANA scale-out deployments, For more information, see Using snapshots with scale-out deployments.
• Your must ensure the following for your /hana/data volume:
  • It is independently hosted on a SSD-based Persistent Disk or Hyperdisk volume with no striping. This disk must host no other SAP HANA volume. As described in the SAP Note 2039883 - FAQ: SAP HANA database and data snapshots (storage snapshots), to be able to recover your SAP HANA database, the storage snapshot must not include the log area of the database.
  • It is mapped to a logical volume. This enables the agent to correctly handle the mount point during disk snapshot based backup or recovery operations.
• If you have deployed your SAP HANA system by using Infrastructure as Code (IaC) tools such as Terraform, then the agent can create snapshot based backups but cannot perform recovery operations. For more information, see Using snapshots with IaC tool based deployments.
• You must perform backup and recovery operations by using the agent instance running on your SAP HANA host VM. You cannot use agent instances running on other VMs.
• To back up and recover your SAP HANA database, you must grant sudo access to certain operating system commands either by providing root access or by using an appropriate sudo group. For more information, see Set permissions.

Required permissions

To allow Google Cloud's Agent for SAP to back up and recover your SAP HANA database by using disk snapshots, you must set permissions for the following:

• The Google Cloud project where your SAP HANA system is running.
• The user that runs the agent commands in the operating system, which could be the SAP HANA operating system user, or your own user.
• The SAP HANA database user used by the agent.

For information about the permissions that you must grant, see Set permissions.

Create disk snapshot based backups

To create disk snapshot based backups for SAP HANA by using Google Cloud's Agent for SAP, you run the hanadiskbackup command. For instructions, see Back up and recover SAP HANA by using a disk snapshot.

The following flowchart shows the steps that the agent performs to create the backup:

If the agent fails to complete any step in this backup procedure, then it marks the snapshot creation as UNSUCCESSFUL and exits the procedure. You then need to manually review and re-run the hanadiskbackup command. You can view the entries in the /var/log/google-cloud-sap-agent/hanadiskbackup.log file to find out why the agent exited the backup procedure.

Create delta backups

When you run the hanadiskbackup command, the agent creates a standard snapshot for the specified disk.

An advantage of standard disk snapshots is that they are incremental, where each subsequent backup only stores incremental block changes instead of creating an entirely new backup. This means that after your create the first disk snapshot for SAP HANA by using Google Cloud's Agent for SAP, all the succeeding snapshots that you create are delta backups by default. You need not take any additional action to create delta backups.

For more information, see How incremental standard snapshots work.

Encrypt disk snapshots

By default, data in Google Cloud is encrypted at rest using AES-256. This applies to the data stored in disks and disk snapshots. Alternatively, you can encrypt your snapshots by using either a customer-managed encryption key (CMEK) or a customer-supplied encryption key (CSEK).

You can also implement an additional layer of security by enabling the encryption provided by SAP HANA for the data volume. For information about this option, see Enabling Encryption of Data and Log Volumes.

Recover SAP HANA data by using disk snapshots

To recover your SAP HANA database by using a disk snapshot based backup, you run the hanadiskrestore command. For instructions, see Back up and recover SAP HANA by using a disk snapshot.

The following flowchart shows how you use the agent's disk snapshot feature to recover an SAP HANA database:

If the agent fails to complete any step in this procedure, then it exits the procedure. You then need to manually review, troubleshoot if required, and then re-run the hanadiskrestore command. You can view the entries in the /var/log/google-cloud-sap-agent/hanadiskrestore.log file to find out why the agent exited the database recovery procedure.

Recover to a specific point-in-time

To recover your SAP HANA database to a specific point in time, in addition to a snapshot of the disk hosting your /hana/data volume, you need the necessary SAP HANA delta and log backups that you previously created for your database.

You can create the SAP HANA log backup either by using the Backint feature of the agent, or a file system backup of your SAP HANA database.

While performing recovery, if you specify a timestamp in the future, then your SAP HANA database is recovered to the most recent state.

Recover from an external source

You can use the disk snapshot based backups that you created for one SAP HANA system to recover another SAP HANA system. You can also use this capability to move systems across environments, for example, from a testing environment to a quality assurance environment.

For instructions, see Perform system copy or system refresh.

Using disk snapshots with SAP HANA deployments

The following sections provide scenario-specific planning information for the backup and recovery of SAP HANA scale-up systems by using the disk snapshot feature of Google Cloud's Agent for SAP:

• Using snapshots with HA deployments
• Using snapshots with DR deployments
• Using snapshots with scale-out deployments
• Using snapshots with IaC tool based deployments

Using snapshots with HA deployments

In an SAP HANA scale-up system running in a high-availability (HA) cluster on Google Cloud, you need to install Google Cloud's Agent for SAP on both the primary node and the standby node in the cluster.

To create disk snapshot based backups for an SAP HANA HA system, you need to run the hanadiskbackup command only on the /hana/data disk in your primary node. If a failover event occurs, then you need to run the hanadiskbackup command on the new primary node. Also, while running the command, you must specify the SAP HANA hostname or IP address of your primary node's host. Don't specify the SAP HANA virtual IP (VIP) used in the cluster.

While performing recovery operations, you can use the disk snapshot to recover the disk hosting the /hana/data volume in either the primary or the standby node. You can recover your SAP HANA database to the time the disk snapshot was created, or any specific point in time.

Using snapshots with DR deployments

For an SAP HANA scale-up system in a disaster recovery (DR) deployment on Google Cloud, you need to install Google Cloud's Agent for SAP on both the primary and standby hosts.

To create disk snapshot based backups for an SAP HANA HA system, you need to run the hanadiskbackup command only on the /hana/data disk in your primary database instance. If you break the SAP HANA system replication and promote the standby database instance, then you need to run the hanadiskbackup command on the new primary instance.

For DR deployments, you perform recovery operations on the primary database instance. You can recover your SAP HANA database to the time the disk snapshot was created, or any specific point in time.

Using snapshots with scale-out deployments

The disk snapshot feature of Google Cloud's Agent for SAP is not compatible with SAP HANA scale-out deployments.

However, to back up and recover an SAP HANA database in a scale-out deployment, you can use the following options:

• Use the Google Cloud Backup and DR service. For more information, see Back up HANA 1+n and HANA scale-out databases.
• Use the Backint feature of Google Cloud's Agent for SAP. For more information, see Backint based backup and recovery for SAP HANA.

Using snapshots with IaC tool based deployments

The disk snapshot feature of Google Cloud's Agent for SAP is partially compatible with SAP HANA systems that have been deployed using Infrastructure as Code (IaC) tools such as Terraform.

While the agent can create snapshot based backups for these systems, the agent cannot perform recovery operations. This is because snapshot based recovery by using the agent includes recreating the disk for the /hana/data volume, which can cause conflicts or issues with how the IaC performs state management.

For example, after you perform recovery by using the agent, whenever your IaC tool checks the status of the disk hosting the /hana/data volume, the tool would try to revert to your system's defined configuration by either reattaching the previous disk, or by deleting the new disk and recreating the previous disk. Any of these actions can severely affect your SAP HANA database, to the extent of making it unusable.

However, most IaC tools provide workflows that let you modify your system's state to keep up with changes to the managed infrastructure. The following is a high-level procedure that you can use to recover SAP HANA systems that you deployed by using IaC tools:

1. Stop your SAP HANA database. For instructions, see the SAP document Starting and Stopping SAP HANA Systems.

2. Stop the Compute Engine VM instance that hosts your SAP HANA system. In an HA or DR deployment, stop all VMs. For information about how to stop a VM, see Stop a VM.

3. In the IaC configuration file that contains the definition of your SAP HANA system, modify the definition of the disk for the /hana/data volume so that it can be recreated by using an existing disk snapshot. For an HA or DR deployment, perform this action for the primary SAP HANA node.

   For example, if you deployed your SAP HANA system by using one of the Terraform configurations provided by Google Cloud, then in the ./terraform/modules/sap_hana/main.tf file, include the snapshot argument as follows:

   # Split data/log/sap disks
   resource "google_compute_disk" "sap_hana_data_disks" {
    count   = var.use_single_shared_data_log_disk ? 0 : var.sap_hana_scaleout_nodes + 1
    name    = format("${var.instance_name}-data%05d", count.index + 1)
    type    = local.final_data_disk_type
    zone    = var.zone
    size    = local.data_pd_size
    snapshot = "hana-SID-snapshot-YYYYMMDD-HHMMSS"
    project = var.project_id
    provisioned_iops = local.final_data_iops
   }
   

4. Apply the modified configuration. For example, if you're using Terraform, then run the following commands:

   terraform validate
   terraform apply
   

5. Recover your SAP HANA database. For instructions, see Recover the database by using a disk snapshot.

Enable logging

While performing disk snapshot based backup and recovery operations, Google Cloud's Agent for SAP writes to the following log files: hanadiskbackup.log and hanadiskrestore.log, respectively.

These log files are located in the /var/log/google-cloud-sap-agent/ directory on your host.

By default, logging is enabled during backup and recovery operations. The default log level is info. You can increase or decrease the log level by using the -loglevel argument when you run the hanadiskbackup and hanadiskrestore commands.

Enable monitoring

By default, Google Cloud's Agent for SAP emits the following metrics to Cloud Monitoring each time you run the hanadiskbackup command. This is controlled by the -send-metrics-to-monitoring argument, which has the default value TRUE.

You can set up monitoring for your SAP HANA database by configuring alerts based on these metrics. You can also view the metrics on a dashboard.

The following table describes the metrics that the agent emits during snapshot based backup and recovery operations:

Metric	Description
workload.googleapis.com/sap/agent/hanadiskbackup/status	Boolean Informs if the disk snapshot based backup operation is successful or not. The value true means success and false means failure.
workload.googleapis.com/sap/agent/hanadiskbackup/totaltime	Int Informs the time, in seconds, that the backup operation takes to create the disk snapshot, including the time to upload it to the specified Cloud Storage bucket.
workload.googleapis.com/sap/agent/hanadiskbackup/dbfreezetime	Int Informs the time, in seconds, that your SAP HANA file system was freezed during the disk snapshot based backup operation.

Best practices

While using disk snapshots to back up and recover your SAP HANA database, we recommend the following best practices:

• Prepare alternative recovery options for your database. We strongly recommend that you design your backup and recovery strategy such that it combines different methods. For example, you can use the agent's disk snapshot feature in combination with a full backup of your SAP HANA database created by using the agent's Backint feature or by using a file system backup.

• Perform consistency checks for your /hana/data volume and disk snapshots. Because block-level integrity checks are not performed during snapshot based backup operations, we recommend that you regularly perform the following consistency checks by using the hdbpersdiag tool:

  • Perform page consistency checks for your /hana/data volume. This activity requires downtime as the tool won't be able to check the data pages of a specific data volume if there is a process using the data volume, such as the SAP HANA index server.
  • On a VM other than your SAP HANA host, deploy a temporary disk from a snapshot that you created by using the agent's hanadiskbackup command, and then perform a consistency check on that disk. For information about how to do this, see Validate snapshot consistency

  For information about the hdbpersdiag tool, see the SAP document Perform Consistency Checks.

• Perform maintenance activities for your disk snapshots. Because Google Cloud's Agent for SAP doesn't perform any maintenance activities for your disk snapshots, we recommend that you periodically review them and perform activities such as deleting old snapshots.

  For more information about disk snapshot related best practices, see Best practices for Compute Engine disk snapshots.