February 18, 2015 at 14:44Tape archiving in Veeam Backup & Replication v8: new for the familiar old
- Tutorial
Many organizations use tape devices in their backup infrastructure. Magnetic tape has been and remains an effective tool for long-term storage of backups, allowing you to implement the rule "3-2-1" data protection. That is why Veeam continues to improve its support for tape libraries, which is what I will talk about today.
For details, welcome to cat.
In the distributed architecture of Veeam Backup & Replication v8, a component with a new role has appeared - this is Tape Server , that is, a server that is responsible for transferring data (backup copies of virtual machines or files) to a tape device. This role is played by the Windows server connected directly to the tape library (via SCSI / FC / SAS) or remotely via iSCSI and acting as a proxy server - in particular, managing read / write, move and load / unload tapes on the tape device. Tape Server
Roleit can be assigned both to a dedicated server, and to a server with an existing role (for example, Veeam backup proxy), or even to a Veeam backup server - for example, if you install "all on one machine". Both physical and virtual machines can cope with this role - the main thing is that it works under Windows and is connected to the tape device, for which it is necessary to install the necessary drivers (for more details, see here in English).
We connect the tape library to the machine that will act as the Tape Server , and on this machine we check in the Device Manager device manager that the tape device is displayed under the Tape drives node :
If, instead of the expected tape device, “Unknown Media Changer" or some other unidentified device appears in the tree, check that the necessary drivers are installed correctly (they must be "native", that is, from the library manufacturer).
Then go in Veeam Backup & Replication Console, open the presentation where the Tape Infrastructure , select the menu command Add Tape Server , and step through the wizard add the Tape Server . In this case, you can select a server that is already in the backup infrastructure from the list, or you can add a new one:
After the steps of the wizard are completed, Veeam will scan the new Tape Serverto determine the presence of the required components and cassettes in the library, and displays in the console a list of available devices that write to the tape ( Drives ), and the tapes ( Media ):
In the previous version, it was required to connect the tape device to the Veeam Backup server, which imposed a number of restrictions - for example, you had to deploy the Veeam server exclusively on a physical machine. Naturally, the load on the server, which was responsible for reading and writing to the tape, was growing, while there was practically no way to optimize traffic.
With the assignment of the Tape Server role to a dedicated machine, the Veeam Backup server is freed from the need to perform read / write operations to the tape and is now exclusively engaged in management, and for Tape Server you can apply traffic optimization settings (see more details here ).
We also note that now it is not necessary to deploy the Veeam Backup server on a physical machine - you can install it on a virtual machine and use the capabilities of VMware Fault Tolerance to protect it.
Among the advantages of using a dedicated Tape Server are the following:
As you can see from the diagram, the tape archive directory (Tape Catalog) is stored in the database on the Veeam Backup server, and if the Tape Server fails, the data on the archives processed by it will not be lost - and the server can be replaced. To backup the database, we use the updated Configuration Backup functionality , which further increases the security of the catalog data.
In addition to the role of Tape Server , support for Tape Vaults was added - these are logical containers for displaying those cassettes that are currently unloaded from the library (offline media). The user can create such containers himself and appropriately group his archives that are currently located outside the tape device (they are in the safe, on the shelf, etc.).
For example, you can create one container for each remote site in the infrastructure. When the cassettes are unloaded from the library and “left” to the place of long-term storage (safe, shelving, etc.), in the Veeam Backup console you just need to move them to the appropriate container. A well-constructed logical structure will allow people responsible for recovering data from tape archives to quickly find the right tapes.
And, of course, we could not help but mention the new algorithm for archiving backups to magnetic tape under the beautiful name Virtual Synthesized Full Backup for Tape. It allows you to synthesize full backups directly on the tape without having the backup job do this on disk first. The archive job (tape job) forms such a complete backup right in the course of its work.
To create this “virtual full backup on tape”, an auxiliary .VSB file (Veeam Synthetic Backup) is used, which contains pointers to data blocks in the files of the backup chain stored in the repository.
When creating an archiving task at the step of the Media Pool wizard, you can click the Schedule button and specify on which days you want the tape to form a full backup using the appropriate chain:
On the day on which this operation is assigned, 3 steps are performed:
The resulting full backup on the tape will be fully suitable for use as a starting point for subsequent increments, which the backup task will already create.
I draw your attention to the fact that such “virtual full backups for tape” can only be created for backup copy jobs and backup jobs that build an infinitely incremental chain — that is, only for chains containing necessarily 1 full backup and a set of subsequent increments.
For details, welcome to cat.
New in Architecture - Veeam Tape Server
In the distributed architecture of Veeam Backup & Replication v8, a component with a new role has appeared - this is Tape Server , that is, a server that is responsible for transferring data (backup copies of virtual machines or files) to a tape device. This role is played by the Windows server connected directly to the tape library (via SCSI / FC / SAS) or remotely via iSCSI and acting as a proxy server - in particular, managing read / write, move and load / unload tapes on the tape device. Tape Server
Roleit can be assigned both to a dedicated server, and to a server with an existing role (for example, Veeam backup proxy), or even to a Veeam backup server - for example, if you install "all on one machine". Both physical and virtual machines can cope with this role - the main thing is that it works under Windows and is connected to the tape device, for which it is necessary to install the necessary drivers (for more details, see here in English).
We connect the tape library to the machine that will act as the Tape Server , and on this machine we check in the Device Manager device manager that the tape device is displayed under the Tape drives node :
If, instead of the expected tape device, “Unknown Media Changer" or some other unidentified device appears in the tree, check that the necessary drivers are installed correctly (they must be "native", that is, from the library manufacturer).
Then go in Veeam Backup & Replication Console, open the presentation where the Tape Infrastructure , select the menu command Add Tape Server , and step through the wizard add the Tape Server . In this case, you can select a server that is already in the backup infrastructure from the list, or you can add a new one:
After the steps of the wizard are completed, Veeam will scan the new Tape Serverto determine the presence of the required components and cassettes in the library, and displays in the console a list of available devices that write to the tape ( Drives ), and the tapes ( Media ):
What are the benefits of such an architecture?
In the previous version, it was required to connect the tape device to the Veeam Backup server, which imposed a number of restrictions - for example, you had to deploy the Veeam server exclusively on a physical machine. Naturally, the load on the server, which was responsible for reading and writing to the tape, was growing, while there was practically no way to optimize traffic.
With the assignment of the Tape Server role to a dedicated machine, the Veeam Backup server is freed from the need to perform read / write operations to the tape and is now exclusively engaged in management, and for Tape Server you can apply traffic optimization settings (see more details here ).
We also note that now it is not necessary to deploy the Veeam Backup server on a physical machine - you can install it on a virtual machine and use the capabilities of VMware Fault Tolerance to protect it.
Among the advantages of using a dedicated Tape Server are the following:
- Tape Server can be located as close to the tape library as possible.
- Depending on the planned load on the Tape Server, you can find the optimal configuration (CPU, memory, bandwidth).
- You can work with several tape devices located in different places, assigning the Tape Server role to several machines and controlling the process from a single console on the Veeam Backup server:
As you can see from the diagram, the tape archive directory (Tape Catalog) is stored in the database on the Veeam Backup server, and if the Tape Server fails, the data on the archives processed by it will not be lost - and the server can be replaced. To backup the database, we use the updated Configuration Backup functionality , which further increases the security of the catalog data.
What else is new in tape support?
In addition to the role of Tape Server , support for Tape Vaults was added - these are logical containers for displaying those cassettes that are currently unloaded from the library (offline media). The user can create such containers himself and appropriately group his archives that are currently located outside the tape device (they are in the safe, on the shelf, etc.).
For example, you can create one container for each remote site in the infrastructure. When the cassettes are unloaded from the library and “left” to the place of long-term storage (safe, shelving, etc.), in the Veeam Backup console you just need to move them to the appropriate container. A well-constructed logical structure will allow people responsible for recovering data from tape archives to quickly find the right tapes.
And, of course, we could not help but mention the new algorithm for archiving backups to magnetic tape under the beautiful name Virtual Synthesized Full Backup for Tape. It allows you to synthesize full backups directly on the tape without having the backup job do this on disk first. The archive job (tape job) forms such a complete backup right in the course of its work.
What is this “synthesis" during archiving?
To create this “virtual full backup on tape”, an auxiliary .VSB file (Veeam Synthetic Backup) is used, which contains pointers to data blocks in the files of the backup chain stored in the repository.
When creating an archiving task at the step of the Media Pool wizard, you can click the Schedule button and specify on which days you want the tape to form a full backup using the appropriate chain:
On the day on which this operation is assigned, 3 steps are performed:
- The archiving task creates a .VSB file and saves it to the repository, next to the backup chain consisting of a full backup (.VBK) and increments (.VIB)
- Checking the list of pointers in this .VSB, the task finds out which blocks (from the .VBK and .VIB files stored on the disk) will be needed to form a complete backup that is relevant for the day. The corresponding blocks are also recorded on the tape in the form of this very “virtual full backup for tape”.
- At the end of the session, the archiving task deletes the irrelevant .VSB file from the repository.
The resulting full backup on the tape will be fully suitable for use as a starting point for subsequent increments, which the backup task will already create.
I draw your attention to the fact that such “virtual full backups for tape” can only be created for backup copy jobs and backup jobs that build an infinitely incremental chain — that is, only for chains containing necessarily 1 full backup and a set of subsequent increments.