Data Recovery - CommCell® Console

A sound data recovery strategy is essential for unplanned events like a system failure, accidental file deletion, system crashes, or larger scale disasters. Commvault software provides different recovery methods depending on the application or file system being protected. This flexibility allows the administrator to pick the optimal recovery method based on the situation to recover data in an efficient manner.

Recovery methods are divided into two main categories:

• Indexed
• Non-Indexed

Indexed-Based Recovery Methods

*Not available in the Commvault Command Center^TM

Non-Indexed-Based Recovery Method

* Not available in the Commvault Command Center^TM

Find

The Find operation is the preferred method for recovering files. Available at the backup set level, the Find operation scans the index database (in the case of V2 indexing), or multiple indexes (in the case of V1 indexing) within a specified range of backup time and searches for a specific filename or pattern (wildcards). You can also limit your scope of search to a specific folder or folder structure. Matching results are displayed within the specified time range. You can select to restore any, all, or specific version(s) of the files within the display.

If multiple versions are restored, each version has a sequential number appended to the filename beginning with 1 for the most recent version of the file.

With email, you can use the Find operation to search message metadata using the 'From,' 'To,' and 'Received' fields of the message.

Browse and Restore - Demo

Browse and Restore

A Browse and Restore operation allows the administrator to browse through the folder structure to select files and folders to restore. You can select multiple files and folders for recovery operations. If a parent object in the folder structure is selected, then all objects within the parent folder are automatically selected for restore.

When selecting a file that was modified multiple times during a cycle, the specific version of the file or all versions can be selected to be recovered.

Image and No-Image Browsing

The following features are used to browse and restore data:

• Image Browsing
• No-Image Browsing

Image Browse

Each time a backup operation is conducted an image file is generated, which represents a view of the folder structure at the time the backup occurred. By default, when a browse and restore operation is conducted, an 'image browse' method is used to present the folder structure as it existed based on the browse date and time. This is done by displaying the folder structure from the most recent image file prior to the point-in-time being browsed. So, if a browse is being conducted on Wednesday at 2:00 PM and the most recent backup was run on Tuesday at 10:00 PM, the image file from the 10:00 PM backup is used. This 'image browse' method produces a consistent structure of the data according to the browse time. This is important since folder structures may change from day-to-day during a cycle.

When restoring an entire folder structure, it is important that the structure represents a specific point when a backup was conducted—and not represent data for the entire cycle. This is best explained by using temporary files as an example. Temporary files and folders can be generated, deleted and regenerated multiple times during a cycle. Each time a backup is run, the file folder structure is different based on which files existed at the specific point-in-time. When a restore operation is run, you would not want every temporary file and folder to be restored, just a particular point-in-time or day.

Although the 'image browse' method is beneficial for restoring file and folder structures to a particular point-in-time, it may also result in deleted items not showing up when a Browse and Restore operation is conducted. For example, if on Wednesday at 2:00 PM a browse operation is run using the Tuesday 10:00 PM image file, and a file or folder was deleted on Tuesday at 2:00 PM, the deleted files will not appear in the browse results. This is because when the 10:00 PM image file was created, the deleted files were not present.

No-Image Browsing

The 'no-image browse' is used to retrieve data that may have been deleted at some unknown time. It browses all the data (including deleted items) for the selected backup set according to the browse time. It is also useful for retrieving a previous backup version and showing deleted files across cycles. 

There are two options to ensure deleted items are displayed during Browse and Restore operations:

• Select the Show Deleted Items check box - This runs a 'no-image browse.' In this case, the image files are bypassed, and the browse operation returns results from the index directory, which shows all items backed up from the point the full was run. This method is useful when recovering user data that has been deleted but may not be a good choice when restoring an entire folder structure, especially if the folder structure was modified during the cycle.
• Specify date and time to browse - This runs 'image browse.' If you know when the data was deleted, specify that date and time in the Browse and Restore options. So, if data was deleted at 2:00 PM on Tuesday, and you specify Tuesday as the browse date, then the most recent image file prior to the point the browse is being conducted would be Monday at 10:00 PM. Since the data was deleted on Tuesday it would be present in the image file on Monday night and will show up in the browse results.

Pie Chart Browse View

The pie chart view in the browse window provides insight into what data is stored on a client and provides a visual method for browsing data. It can be useful when looking for a certain type of file.

To display the pie chart, launch a traditional Browse and Restore, then click the pie chart icon in the upper-right corner.

Restore Options and Tasks

Whether the Browse and Restore or the Find option was used to locate data, the next step in the restore process is to recover the data. There are several options available for restoring data depending on the agent type and the operating system.

Key points when restoring data:

• Data can be restored in-place, which is on the same system, in the same location.
• Data can be restored out-of-place, on the same server in an alternate location or on a different server.
• A file name suffix can be appended to restored file names for ease of identification.
• Files can be restored in a remote location by providing a UNC path and the user who has access to the share.
• The List Media option is used to ensure the required media are in the library or if it must be recalled.

Basic Recovery Options

Select data to restore | Click Recover All Selected | General Tab

Multi-Stream Restore

Use Browse or Find to select objects to restore | Click Recover All Selected | General tab

Commvault^® Version 11 Service Pack 6 introduces the ability to uses multiple streams during an index-based restore such as Browse and Restore or Find. By default, the system traditionally uses one stream, but this number can be increased by using the 'Number of streams' option in the Restore Options General tab. Multi-stream restore is only supported for data using V2 indexing.

Copy Precedence

Click Browse or Find | Choose Advanced Options tab

Each storage policy copy within a storage policy has a copy precedence number assigned to it. By default, the primary copy has a precedence number of one. The next secondary copy created will have a precedence number increased by one in the order in which they are created.

When restoring data using Commvault^® software, by default, it restores the data from the lowest copy precedence number where the data is available. For example, data is stored on disks for 30 days and copied on tapes for 90 days. If you browse for data that is earlier than 30 days, there is no need to specify the disk copy. The system automatically goes to the disk copy since the primary copy has a precedence number one (1). If you browse for data older than 30 days, the system automatically goes to the tape copy, which has a precedence number two (2).

If it is required to restore data from a higher precedence number (even if the data is available from a lower precedence copy), you can override the default behavior by specifying a precedence number in the Browse and Restore or Find operations' Advanced Options tab.

Tip: Validate Restores from a Tape Copy
Scenario: You are backing up data to disks and copying the data to a secondary tape copy to send offsite daily. Your manager is asking you to prove that you can restore data from tapes before sending it to the offsite location.

Solution: To restore data from tapes and avoid the system automatically restoring from disks, specify the tape copy precedence number during the restore.

Append a Suffix to Restored Files

Select data to restore | Click Recover All Selected | Advanced button | Map tab

During a restore it is possible to append a suffix to restored files. It is useful if you wish to restore the files in the same location of the production files without overwriting them. The suffix is appended at the end of the file name, right before the file extension.

Apply Filters to a Restore

Select data to restore | Click Recover All Selected | Advanced button | Paths/Filters tab

In some cases, it is useful to apply filters during a restore. For instance, if you are restoring folders that belong to several users, but would like to filter out AVIs, simply apply a filter to the restore. Filters are defined in the Paths/Filters tab of the Restore Advanced options.

Run Scripts Before and/or After the Restore

Select data to restore | Click Recover All Selected | Advanced button | Pre/Post tab

During a restore it can be helpful to run a script either before or after the restore completes. For instance, if you want to launch an application or start a Windows^® service, Commvault^® software allows the script to be executed from the Pre/Post tab of the restore job's Advanced Options.
To use scripts before and/or after restoring, copy the scripts onto the client server.

Additional Recovery Methods

Restore by Job

The Restore by jobs operation is a 'non-indexed restore' that uses multiple concurrent streams during the restore operation. Unlike a traditional Browse and Recovery operation – which requires index files to be available in the index directory and uses a single stream during the restore operation, the 'Restore by Jobs' provides a faster recovery in DR situations when the production index directory location is not available. Since this operation does not require indexes and it is a chunk-based restore that allows multiple streams to run simultaneously, restore operations run immediately and provides a faster restore speed.

There is a negative aspect of the Restore by Jobs operation. The single pass restore method used with indexed-based recoveries (where only the proper version of an object is restored based on the point of browse) is not used. This means that in order to bring a machine back to its last state, the last full backup job would need to be recovered first followed by each subsequent incremental job ending with the most recent.
The Restore by Job option restores an entire job and therefore does not offer any restore granularity.

Full System Restore

The Full System Restore or 'full agent restore' operation is an indexed-based restore where the entire backup set is selected when browsing for data. The backup set selection includes the entire contents of the file system backup, including configuration information such as system state. This is a restore type that can be used in case of a full system crash.

For this to work, several steps must be accomplished:

1. Build a new machine with similar hardware and same mass storage configuration.
2. Install a working Operating System of the same level as the crashed client (i.e. Windows 2008R2 SP2).
3. Install the client agent on the target system, but configured using the crashed system client name.
4. A full backup of the system including the Windows^® system state must be available. The system state backup should not have critical components filtered. Here are the required critical system state components:
   • Registry
   • System Protected Files
   • Active Directory
   • Cluster DB
   • Sysvol (For 2000 or 2003 domain controllers level)
   • DFSR (Windows 2008 Domain controller level)
5. Browse at the backup set and select the entire backup set.
6. Perform the restore.
7. Reboot.

Restore Validation

Despite the high availability and redundancy of IT infrastructures, in some cases, it may be required to validate if data is restorable. It can be an easy task if the available storage can accommodate a test restore. But it quickly becomes a challenge when available space is limited.

The 'Validate Only' restore option, available for several agents such as file system and SQL, helps to address that challenge by validating the restore without actually writing any data on the client. The restore process proceeds as usual; reading the data from the MediaAgent library and sending the data to the client. When the client receives the data, it validates if the received data size is the same as the size in the headers. The client then discards the data without writing it anywhere. The job details provide metrics for the restore, making the 'Validate Only' option a great tool for testing network performance.