LAN-Free Backup Procedure
This section contains general operational issues and a procedure for performing a LAN-free backup.
General Operational Issues
You will need to establish general operational procedures, automated or otherwise, to employ the point-in-time copy technology. You will also need to decide when the backup snapshots occur, how often they occur, and how the backup software is notified when the snapshot is available.
In most cases, a point-of-data-consistency needs to be reached before a point-in-time can be established. This ensures that the snapshot image is logically consistent (that is, the application can recover from any incomplete transactions).
The following list contains the general steps for creating a point-in-time copy that can then be accessed by a test application.
Make sure that the master application has reached a point of consistency (that is, it has flushed all data to the disk).
Create a PIT copy of the data.
Export the PIT copy to a backup server by using the deport shadow capability.
Initiate tape backup.
When required, resynchronize the PIT copy with the master volume data by using the Rejoin and the Fast Resynchronize functions.
To Perform a LAN-Free Backup
This procedure describes how to make a PIT copy and export that copy over the SAN to a backup server. The following steps assume that the Instant Image 3.0 software is installed on all of the computers that have access to either the master or snapshot image.
Log in to the system as superuser.
Determine which volumes are the master volumes and which volumes you will use for the independent shadow volumes.
It is key that the shadow volumes be on SAN-attached or dual-ported storage so that other servers can access these volumes.
This step typically requires some planning. The idea is that the master volumes and original bitmaps remain available to the application server at all times. The PIT copy and its newly created secondary bitmaps should be available to the second server or backup server. Therefore, put the master and bitmap volumes in a diskgroup separate from the shadow and secondary bitmap volumes. This enables the deported shadow and its associated secondary bitmap volume to be deported and imported as a single diskgroup.
Calculate the maximum amount of raw volume to be duplicated.
For an independent copy, you need a duplicate amount of usable storage to store the point-in-time copy. Take your existing diskgroup and volumes and duplicate them to another diskgroup. The online or master volumes can be any RAID level supported by the Solaris OE or by a volume manager. The PIT, or shadow volumes, does not have to be the same RAID level as the master. The independent shadow volumes must be at least as large as the master volumes.
Make both the master and shadow volumes available for the Instant Image software.
Create two diskgroups, one for the online production volumes and another for the snapshots.
Any volume that is available to the Solaris OE is eligible to be used with the Instant Image software. If the shadow volume is to be exported to a second server, the shadow volumes must be in their own diskgroup. This enables the shadow disk group to be exported as a single entity using VxVM.
Create a volume for the bitmaps that can be export and imported along with the snapshot.
The easiest way to do this is to create a volume within the diskgroup.
Use VxVM or the format(1M) command to allocate space for the Instant Image software to keep track of the differences between the master and the shadow volumes.
This is the bitmap associated with the master volume. The Instant Image software requires approximately 8 Kbytes for each gigabyte of raw disk space, plus space for header information (24 Kbytes).
One bitmap is required for each Instant Image volume group (a group contains one master and one shadow volume). It is recommend that raw volumes be used for the bitmap volumes.
The following is an example of the bitmap volumes used in this document:
/dev/vx/rdsk/production/vol01.bmp /dev/vx/rdsk/production/vol02.bmp /dev/vx/rdsk/production/vol03.bmp /dev/vx/rdsk/production/vol04.bmp /dev/vx/rdsk/production/vol05.bmp
Additionally, the same bitmap volumes should be created in the backup diskgroup. It is these bitmaps that are used on the secondary server.
Prepare the application and/or file system to be backed up.
Any data that is in the server page cache should be flushed to disk. For the UNIX_ file system (UFS), use the lockfs(1M) command. Databases have different procedures for flushing the page cache. For example, The Oracle_ database software uses the SQL commands BEGIN and END BACKUP. The Informix database software uses its onmode command. No matter which method you use, get all of the data to be backed up to the disk.
Enable the point-in-time volume sets.
For manageability, all of the volumes that belong to this server-application should be grouped in an Instant Image software I/O group.
This enables you to manage this backup as a single set.
# iiadm -g Backup1 -e ind /dev/vx/rdsk/production/vol01 /dev/vx/rdsk/ backup/vol01 /dev/vx/rdsk/production/vol01.bmp # iiadm -g Backup1 -e ind /dev/vx/rdsk/production/vol02 /dev/vx/rdsk/ backup/vol02 /dev/vx/rdsk/production/vol02.bmp # iiadm -g Backup1 -e ind /dev/vx/rdsk/production/vol03 /dev/vx/rdsk/ backup/vol03 /dev/vx/rdsk/production/vol03.bmp # iiadm -g Backup1 -e ind /dev/vx/rdsk/production/vol04 /dev/vx/rdsk/ backup/vol04 /dev/vx/rdsk/production/vol04.bmp # iiadm -g Backup1 -e ind /dev/vx/rdsk/production/vol05 /dev/vx/rdsk/ backup/vol05 /dev/vx/rdsk/production/vol05.bmp
The above commands create an Instant Image software I/O group called Backup1 and inserts each master and shadow volume set into that group. You can now manage the snapshot as a single entity. Although, you have enabled all of the volume sets into the same I/O group, at this time, they all have different PITs. To set a single PIT, you must invoke an Instant Image software update operation.
The following is an example of an update operation.
# iiadm -g Backup1 -u
In versions 1.x and 2.x of the Instant Image software, it was necessary to register the volumes into the storage volume software. This is no longer necessary. The iiadm -e command performs all of the needed storage volume registrations.
Bring the application back online, if necessary.
In most cases, this step is not necessary because the application was never taken offline. For databases, you must take the database out of backup mode, clear any onmode command, and unlock any file systems, if previously locked.
Export the shadow to another server.
After the initial copy has been completed (only done once), you can export the shadow volumes to a secondary server. You can monitor the enable progress by using the iiadm -i command. If you are using scripts, the iiadm -g Backup1 -w command enables the script to wait until the copy is completely established.
The following is an example of how to export the shadow copy:
# iiadm -g Backup1 -E or # iiadm -E shadow_volume (for each volume) # svadm -d shadow_volumes # vxdg -g backup stopall # vxdg -g backup deport (only necessary if using VxVM)
It is necessary to disable the volumes from the storage volume software when they are going to be deported. This is volume manager specific. If no volume manager is used, removing the volumes from storage volume may not be necessary, but it is a good practice to follow. Logical volume managers require that volumes to be deported have no open processes.
Import the shadow disk group and snapshot volumes to the backup server.
Mount the file systems, if necessary. The shadow disk group and snapshot are imported by using the following commands:
# vxdg -g backup import # vxdg -g backup startall # iiadm -I /dev/vx/rdsk/backup/vol01 /dev/vx/rdsk/backup/vol01.bmp # iiadm -I /dev/vx/rdsk/backup/vol02 /dev/vx/rdsk/backup/vol02.bmp # iiadm -I /dev/vx/rdsk/backup/vol03 /dev/vx/rdsk/backup/vol03.bmp # iiadm -I /dev/vx/rdsk/backup/vol04 /dev/vx/rdsk/backup/vol01.bmp # iiadm -I /dev/vx/rdsk/backup/vol05 /dev/vx/rdsk/backup/vol05.bmp # svadm -e shadow_volumes
The next step is to mount the file systems, if any, that reside on the above volumes. The snapshot volumes can be put back into the Instant Image software I/O group, if desired, by using the following command:
# iiadm -g Backup1 -m shadow-1,shadow-2, &.. shadow-n
It is necessary to insert the volumes into the storage volume software to track all of the changes that occur to the shadow volumes while they are being used by the secondary server.
Start the backup process.
The only difference from a normal backup is that the data is now local, instead of out on the network. After the backups have successfully completed, the PIT can be removed from the backup server by using the following commands:
# iiadm -g Backup1 -d # svadm -d shadow_volumes # vxdg -g backup stopall # vxdg -g backup deport startall
Rejoin the master and shadow volumes.
The Instant Image software logs the changes that are occurring on the master volumes as well as the shadow volume.
When you are ready to update the snapshot, export the bitmap from the second server back to the where the master database resides.
The Instant Image software determines which blocks need to be updated. This is referred to as a join. You do not need to recopy the entire volumes to update the shadow volumes. Only the modified data needs to be copied.
Next, ready the snapshot to be brought back to the master.
On the secondary server, stop the application, remove the volumes from storage volume, and deport the volumes by using the volume manger, as in Step 9.
To reattach the snapshot to the master, use the following commands on the master:
# vxdg -g backup import # vxdg -g backup startall # iiadm -J /dev/vx/rdsk/backup/vol01 /dev/vx/rdsk/backup/vol01.bmp # iiadm -J /dev/vx/rdsk/backup/vol02 /dev/vx/rdsk/backup/vol02.bmp # iiadm -J /dev/vx/rdsk/backup/vol03 /dev/vx/rdsk/backup/vol03.bmp # iiadm -J /dev/vx/rdsk/backup/vol04 /dev/vx/rdsk/backup/vol04.bmp # iiadm -J /dev/vx/rdsk/backup/vol05 /dev/vx/rdsk/backup/vol05.bmp # svadm -e shadow_volumes
The above commands take the changes on the secondary server, recorded in the bitmap volumes, and merge them with the changes done on the master-production server.
Resynchronize the master and shadow volumes.
Put the application into a known state (see Step 6).
Execute the following commands to update the shadow volumes:
# iiadm -g Backup1 -u s (for an atomic case) or # iiadm -u s /dev/vx/rdsk/backup/vol01 # iiadm -u s /dev/vx/rdsk/backup/vol02 # iiadm -u s /dev/vx/rdsk/backup/vol03 # iiadm -u s /dev/vx/rdsk/backup/vol04 # iiadm -u s /dev/vx/rdsk/backup/vol05
If this procedure executes from a script file, the Instant Image software can wait for the updates to complete before allowing any other commands. Use the iiadm -w command in the script.
You can apply the same scheme to applications that are on file systems; the Instant Image software works at the raw device level. Therefore, if the application resides on a file system, follow the these steps:
Quiesce the application (put the application into a known state).
Flush the page cache, if necessary.
For example, use lockfs /mount_point on UFS filesystems
Establish or update the PIT copies, as described above.
Unlock the file systems, if previously locked.
Use lockfs -u /mount_point.
Mount and export the backup snapshot.
If you add more volumes and partitions in the future, follow the steps described in this document to enable the Instant Image software. Adding table spaces and data files to volumes that are already being used with the Instant Image software does not require changes to the Instant Image software.
You must modify the preceding for each particular installation. For repetition and ease-of-use, the sample commands used in this document should be put into a script. It is highly recommended that error checking be inserted as well. If an error should occur, the script should stop, and the error should be addressed.