Forum Discussion

Lok_Koo's avatar
Lok_Koo
Level 3
11 years ago

Netbackup Licensing Requirement

Hi All,

 

Sorry but it was a bit tough to find a precise answer suitable for my situation.

 

I have 5 servers in my environment, 4 of them are production servers and are all connected to a SAN storage, another 1 being a UAT server.

I am planning to have 2 of the servers to be master servers, they will be connecting to a tape library with two tape drives. One server will be connecting both tape drives with SAS cables and when it fails we will manually switch the SAS cables to connect to the secondary server.

The UAT server will have an internal tape drive and will be a master server as well.

Does this mean I will need 3 x master server license, 5 client license? Is there a license requirement for the number of tape drives connected?

 

Sorry for the questions and thank you very much for your time.

 

  • Hi,

     

    You license servers, be it the master or a media server, with a server license. This license allows you to write backups to backups storage (be it tape or disk).

     

    For tape drives in libraries you need a library based tape drive. You don't need a license for a internal tape drive.

     

    For clients that write to a server accross the lan you need a standard client.

     

    You can also just license based on your front end data size (the size of all the data on your clients in TB). Then you get most licenses incl.

     

    Ask your partner about licensing.

     

    Your idea to have 3 masters is not very inline with how we do things in NetBackup though. We usually deploy 1 master (in disaster recovery sites maybe another) to control the entire environment. In your environment that would be

    1 server is the master and 4 client,

    or maybe

    1 master, and 1 media (UAT), and the 3 clients.

     

    Your idea to move the tape libary around is possible but a bit of a hassle as you'll need to change the device configuration.

     

    Your solution sounds more like something one would see in Backup Exec.

  • Hi,

     

    You license servers, be it the master or a media server, with a server license. This license allows you to write backups to backups storage (be it tape or disk).

     

    For tape drives in libraries you need a library based tape drive. You don't need a license for a internal tape drive.

     

    For clients that write to a server accross the lan you need a standard client.

     

    You can also just license based on your front end data size (the size of all the data on your clients in TB). Then you get most licenses incl.

     

    Ask your partner about licensing.

     

    Your idea to have 3 masters is not very inline with how we do things in NetBackup though. We usually deploy 1 master (in disaster recovery sites maybe another) to control the entire environment. In your environment that would be

    1 server is the master and 4 client,

    or maybe

    1 master, and 1 media (UAT), and the 3 clients.

     

    Your idea to move the tape libary around is possible but a bit of a hassle as you'll need to change the device configuration.

     

    Your solution sounds more like something one would see in Backup Exec.

  •  

    Thank you for your clarification.

    Though, could you kindly advise a HA solution in this case?

    The idea is to allow the secondary server to continue backing up after the primary server failed (i.e. power loss).

    If I connecting both servers to the tape library at the start, can netbackup be configured in active-standby mode? Will I still need 2 licenses in this case?

    Thank you for your time.

  • Hi,

     

    NetBackup can be installed in an HA cluster but tape devices are not supported on clustered master servers. So that wont work. What you can do is the following

     

    Make 1 server a master, add two media servers to that domain, then share the tape libary between the two media servers.

    1 of the media servers would have control of the robot and the drives would be shared drives (you need an SSO license for this for each tape drive in addition to the library based tape drive license). Should the one fail that controls the robot you can run a reconfiguration and update the robot control and continue.

     

    In this configuration the master (and clients) won't perform backups to the tape drives directly but via the media servers.

  • Hi Riaan,

     

    Thank you for replying.

    Please see whether its possible if I revise my design as follows:

    5 servers, server A to E

    Server A and B will form a cluster, with Master server installed in the cluster for failover (1 Master Server License)

    Server A and B will also each be a media server (may I know is this allowed? or do I have to separate Master Server and Media Server?) (2 media server license?)

    Server A will be have the initial conncetion between the tape drive and itself. If it fails, will manually failover SAS connection to Server B and reconfigure. (2 library base tape drive licenses + 2 SSO license)

    Server C D will be clients (2 x client license)

    Server E (UAT in a different VLAN), will have its own internal tape drive. Configure as media server and store its data in the master server in AB cluster. (1 media server license)

     

    Thank you very much for your time.

     

     

     

  • Hi,

     

    Why do you want a cluster? To have HA for the master server (scheduling, access to the catalog), or HA for the tape drives?

     

  • Hi Riaan,

    The main objective is to provide HA for the backup jobs.

    So in this case we need two servers that are capable to performing backups on its own and two paths from tape drives to those servers so in case one server is down, another one can take over.

     

    Thank you for your time!

  • Hi,

     

    Ok the problem here is that this type of configuration is very enterprise (for large scale datacenters). in NetBackup we don't ever cluster the media server function, in other words the component that writes to tape. That means you won't actually install the software into MSCS and cluster it with a virtual name and that name (IP) then fails over to the other node in case of failure. This is basically because of the nature of SCSI and tape drives.

     

    What we do cluster is the master server component, so if you want to ensure the brains of your NetBackup environment needs to be clustered you can do it. In this type of configuration you'll typically have a 2 node master server cluster, and then media servers on different servers that are connected to the tape devices. The actual master nodes are NOT connected to the tape devices. (if you're using disk storage you can add it to a master server cluster).

     

    So for the media servers you can share the tape drives between them and have storage unit groups that will send data either media server, or you can configure the drives on both, and only use one until the other fails, and then update your policies with a different storage unit (pointing to the new media server).

     

    There is also another concept called a virtual storage unit. This is when you're already running a cluster for say SQL for instance. You then install the media server software on both SQL nodes, and we create storage units for each server. We then create a logical / virtual storage unit based on the virtual server name of the SQL (the name that clients access SQL through). This allows clients (be it the SQL server itself, or clients on the LAN) to send their data to the virtual name, which then maps down to the active physical server with its tape devices.

    I would say put the master unclustered on UAT server and install media servers on A & B giving both access to the tape drives. you can then either use both at the same time or keep one active and one passive.

     

     

  • Hi Riaan,

     

    Thank you for your detailed explanation, really appreciate it.

    However, can the media server still perform the backup if the master server fails?

    If the master server really failed then wouldn't it be single point of failure?

     

    Thank you for your time and patience.

     

  • Correct, if the master fails you'll have no backups. It just seems like overkill in such a small environment to cluster the master. You would need 2 nodes for that, and another 2 nodes for media servers. If you can use virtual servers then sure, go ahead and cluster the master using virtual. Please note tape devices are not supported in virtual so you can't do it for the media servers.