In Tamino, an external backup is a backup that uses another system outside of the database, e.g. software supported special storage devices, to save the content of a database.
The following topics are covered in this document:
General Considerations about External Backups (EMC and Network Appliance)
Configuring External Backups with EMC or Network Appliance Systems
Tamino supports several external backup methods:
Mirroring with EMC systems
Mirroring with Network Appliance systems
Snapshots with Network Appliance systems
The backup method you choose must be compatible with the used hardware. You may use all backup methods on the same computer, but for a given database it is allowed to use only either EMC or Network Appliance. If you have stored your database on Network Appliance, you can use Network Appliance mirroring and Network Appliance snapshots for the same database.
It is not possible to have more than one database on a logical volume. The reason is that the external backup with EMC or Network Appliance saves complete logical volumes, which requires a database synchronization to be done beforehand. These two activities are performed when you invoke the external backup. If a second database was stored on the same volume, a database synchronization (plus the necessary update of the internal configuration store) for that second database would not be done.
The internal configuration store and the log spaces must not be stored on the same logical volume as the database spaces. Otherwise the information on the log spaces or the log spaces themselves would be lost after a restore, without a recovery being possible.
If you restore an old Network Appliance snapshot, all newer snapshots are lost.
If you use snapshots, you can get a database space overflow although you do not add new data to the database, but perform updates only. This is due to the fact that the updates are not performed in the same place, but are copied to wherever there is some free space available. As a consequence, you should monitor the free space on the logical volumes which contain database spaces.
After the first full backup has been done with an external backup system, the backups that follow are done incrementally - only the changes are backed up. This leads to a much better performance of backup and restore operations.
If you run a database on a Network Appliance Filer or EMC Symmetrix, it seems that the database files are part of the file system. On a Windows system, the device has to be mapped to a device letter. On a UNIX machine, the device has to be mounted. In the case of Network Appliance, the filer is divided into volumes. In the case of EMC Timefinder, the Symmetrix with the related disks is divided into device groups. Generally, it is not allowed to run more than one database on a volume or device group. The database should not be on the same volume as $SAG. Additionally, another volume/device group has to be configured for the log location and log archive location. This is very important if Tamino is to run error-free.
As described in the following section Steps, the configuration of the respective external backup method is done within the Tamino Manager. It is possible to configure more than one external backup method. Also, it is possible to use a mix of internal and external backups.
Please note that Tamino does not check if the Database Administrator has configured the external backup correctly. Forgetting a logical volume containing database spaces for the database to be saved would be fatal, since after a restore the database would be inconsistent, due to the fact that some database spaces were not restored.
This section describes the steps required for doing external backups on a Network Appliance Filer or an EMC Symmetrix System. The following topics are covered:
To enable one of the methods, you first need to select the option under External Backup Environment in the Tamino Manager.
For an external backup with Network Appliance Filer, the database
administrator has to allow access to the file system via the command
exportfs. The Network Appliance Filer concept allows
adding and removing disks without stopping any application. If there is not
enough disk space, an additional disk can be plugged in without shutting down
the server or waiting for transactions to finish. This is very important for
24/7 availability. All Network Appliance specific commands produce information
that is logged in an error file. If an error occurs, the error file can be
found in the database specific log location.
There are two backup methods with a Network Appliance Filer: Snapshots and Mirroring.
For this method, you first have to configure a volume for the database files, with a log location that points to a different location. Then, in the Tamino Manager, choose the database you want to back up. Under , choose . To set the filer and volume name, select the option . This opens a dialog that lets you specify the filer and volume name as parameters for the backup method.
After having modified the method parameters, select the object . Then choose from the context menu. In the resulting dialog, choose the desired backup method Network Appliance Snapshot.
Snapshots are available via their name, so the next step is to define it. Choose to display the page Create Backup - Specify Backup Configuration Name and enter a Backup Configuration Name.
Choose to start the backup. During the backup process, the database suspends update requests.
The restore functionality is available as for internal backups (see section Restore and Recover). There are no special requirements for restoring snapshots. Note that it is possible to take more than one snapshot. If you do not restore the latest snapshot, newer backups are also not available any longer, since the snapshots are stored on the same volume. If the whole volume is restored, the snapshots are lost.
For this method, you have to configure a volume for the database, and one volume for each mirrored database. The mirror volumes do not have to be on the same filer, but must have exactly the same sizes as the database volume. The mirror volume must be initialized by the filer administrator. To do so, select the object Network Appliance Mirror under External Backup Environment and choose from the context menu. In the resulting dialog, enter the names of the filer and volume.
The next step is to define a configuration name. To do so, choose and enter a name in the field Configuration Name.
After having configured the parameters for the Network Appliance Mirror, select the object . Then choose from the context menu. In the dialog box that appears, choose the desired backup method Network Appliance Mirror. Then choose and select the backup configuration:
If you choose a mirror of an existing backup, it is overwritten. Consequently, only one backup with a mirror name is managed. Before the system performs the backup, a test is done to find out whether the destination filer for the mirror exists. During the backup process, the database suspends update requests. For the restore process, all other mirror backups are still available. If Network Appliance snapshots are combined with mirrors in the backup sequence, all snapshots later than the restored mirror are no longer available.
Tamino supports database backups in conjunction with EMC's Timefinder software. The concept is to split a BCV (Business Continuance Volume) from the so called standard device(s) where the database is located for usage as a backup save set. The following configurations are supported:
Single BCV on the same Symmetrix
Multi BCV on the same Symmetrix
Remote BCV on a different Symmetrix
The first step you need to take is to configure the external backup environment. Specify the devices where the database is located.
To do so, select the object EMC TimeFinder under External Backup Environment and choose from the context menu. In the resulting dialog, enter the following information:
The target (the Symmetrix ID of the database).
The Symmetrix devices where the database is located.
The library path of the EMC Symmetrix API.
As the database files are located on file systems, you have to
determine on which physical devices the file system is mounted. A file system
may spread over multiple physical disks, particularly if logical volume
managers are involved. The SYMCLI call symhostfs may
help to collect that information.
The following rules must be followed when configuring your database for external backup with EMC Timefinder:
All index and data locations must be taken into account (including the Reserved Location, because a database may open container files at this location).
It is not necessary to take journal and temporary files into account (which means that these files may be located at the same place as index and data). To speed up the backup/restore process, however, it may be more efficient to put the journal in a different location.
The log location and log archive location must be at a different place, which is not part of the external backup process.
You can choose between two kinds of device specifications. Either specify the physical device name (/dev/rdsk…), or the Symmetrix device name (3-hexdigits value):
Physical:
Target = pd Devices=/def/rdsk/... ...
Symmetrix:
Target = <SymmID> Devices=XXX...
The Symmetrix ID (SymmId) can be abbreviated from right to left until it is unique for a host. However, it must have at least 4 digits.
To connect to EMC's software component SYMAPI, you have to define the full path name of the symapi library within the associated EMC installation.
After having defined the standard devices, choose from the context menu to configure the BCV. Note that the number and size of the BCV devices must match the ones of the database. Each BCV receives a freely selectable group name, which is to be used when the backup is done. For a Multi-BCV environment, each BCV set must be entered in the same way.
After having finished the configuration, the database can be backed up to the configured BCV(s). The configuration is checked and the devices are inserted into a device group which receives the name of the database. Thus you can view this device group with the appropriate SYMCLI functions.
There are two phases, the establish and the split phase. During the establish phase, normal database activity is possible without interference. During the split phase, which is less time consuming, the database suspends update requests. Only retrieval is possible.
The first backup after configuration is a full backup. After this first backup, all following backups are incremental. Thus the complete backup and restore process is much faster.
The restore operation restores the BCV and then plays back the redo log. As it is an incremental restore (only changed data is restored), it is a relatively fast action. Because the disks are changed below the mounted file system, it must be remounted (on UNIX) to get rid of cached data. During this restore procedure, all involved file systems will be unmounted. After the restore is completed, they will be mounted again. Thus no other processes must be active on the file systems involved (the mount and unmount processes are managed by Tamino). Note that the BCV-restore plays back the entire file system(s), not only the database files. Therefore you must be aware of the fact that this affects all other files located in such file system(s).
Some basic functions (e.g. completion of a backup or a restore) are logged with EMC's standard logging facility (which is the logfile in the log directory of the EMC installation). If an error occurs, Tamino displays an error message. If a SYMAPI function fails, a detailed error code is displayed. The errors are also logged in the standard EMC log file.
Tivoli Storage Manager provides backup facilities for networks of computers. If you want to include Tamino databases into your Tivoli Storage Manager based backup strategy, you have the following alternatives:
Shut down the Tamino database servers before the backup, back up the data and index spaces using Tivoli Storage Manager, then restart the database servers. In this case, it is recommended to have just one database per location, and to back up all locations assigned to index or data spaces to be sure to include all relevant files.
Switch the database to read-only by changing the server property, then stop and restart the database server. Back up the data and index spaces using Tivoli Storage Manager, switch the read-only property off, then stop and restart the database servers. In this case, it is recommended to have just one database per location, and to back up all locations assigned to index or data spaces to be sure to include all relevant files.
Do a Tamino backup as usual. Back up the resulting backup space with Tivoli Storage Manager.
On UNIX, do a Tamino backup to a named pipe location, using the Tamino database as usual. Have Tivoli Storage Manager read from the named pipe.
The file names of the backup spaces can be displayed with the argbatch
command show backupfiles.