To add a media manager storage unit for the media server nbumedia1with a label of nbumedia1_stu
# bpstuadd -label nbumedia1_stu -host nbumedia1 -density hcart2 -rt TLD -rn 0 -maxmpx 4
-cj 16 -odo 1
The options passed to the bpstuadd command are as follows :
-label : The name of the storage unit
-host : The name of the media server which will use the storage unit
-density : The storage unit will only use tape drives which are configured with the same density and which the media server
has visibility of. If the media server is also configured with tape drives which have a density of hcart3 for example then they could not be used by this storage unit
-rt : (Robot Type) The robot type which houses the drives to be used. This must match the robot type which is hosting the drives to be used.
-rn : (Robot Number)The robot number which is hosting the drives to be used. When a robot is configured a number is assigned to it and that is how the robot is subsequently identified within NetBackup.
- maxmpx : (Maximum Multiplexing) This settings determines the maximum multiplexing factor of the data written to each individual physical tape drive. Multiplexing allows multiple concurrent streams to be written to a single physical tape drive. A tape cartridge, or a VTL configured tape drive, is a serially accessed medium which means only a single stream of data can be written to the magnetic or virtual tape at any one time. A MPX setting of greater than 1 allows a NetBackup media server to send multiple streams of data to the tape drive concurrently. The aim is to reduce the amount of idle time the physical drive encounters when writing data to the cartridge in order to reduce the occurrences of shoe shining. Shoe shining happens when the tape drive has to stop and rewind back to the last end of file marker because data to be written to the tape has stopped arriving. Such a situation is detrimental to the performance of the tape drive and subsequently the backup performance. MPX settings do not apply to physical disk backup targets as disks can be written to with parallel streams. In this example a figure of 4 is given which means 4 concurrent streams could be sent to this tape drive concurrently. Each stream will be written to the tape drive's buffer and the tape drive will be responsible for writing this data to the tape cartridge. The media server is responsible for keeping track of which client data is written where on the physical tape.
-odo : On Demand Only means that the storage unit will only be used in backup policies where it is explicitly referenced in the policy residence field. If "Any Available" is specified as the residency within a policy then this storage unit would not be available due to this option being set to 1.
-cj : (Concurrent Jobs) The name of this setting is something of a misnomer as it relates to the number of drives which can be members of the storage unit. In the example above a figure of 16 is given which means up to 16 drives can be used in this storage unit. With an MPX of 4 and a CJ of 16, this storage unit would be able write 48 concurrent streams of data. Each concurrent stream relates to a single backup job, hence these settings mean that the storage unit group could cater for 48 concurrent individual client backups. If there are less drives available than the number given to the CJ setting then the media server will use up to the number of drives available.
A word about Max Fragment Size : There is no need to set a value for max fragment size, MFS, when configuring a media manager storage unit. The max fragment size was designed to speed up accessing files on tape when tape technology was a lot slower than it is today. With the performance characteristics of modern tape drives the max fragment size can be left at its default value of 1TB.
Storage Unit Groups
A storage unit group allows multiple storage units to be grouped together to provide a level of redundancy in the event that a media server or tape drives go offline. A selection method, given when configuring the storage unit group, determines if all of the storage units in the group will be used in a round robin fashion, load balanced or if the group will be a failover group. If a failover selection method is chosen then in the event of the media server controlling the first storage unit listed in the group, going offline, the backups jobs will be directed to the next storage unit listed in the group and so on. Once the primary media server is back online then jobs will directed at the first storage unit listed in the group again.
Adding a failover Storage unit group
# bpstuadd -group MEDIA_SRV_1_STU1 MEDIA_SRV_2_STU2 MEDIA_SRV_3 -sm 3
-sm : The selection method option determines the type of storage unit group
Additional commands are available for listing, modifying and deleting storage units and storage unit groups
# bpstulist : list only storage units
# bpstulist -g : list both storage units and storage unit groups
# bpstulist -go : only list storage unit groups
# bpstulist -label <storage_unit_name> -U : list details of the named storage unit
# bpstulist -group <storage_group_name> -U : list details of the named storage unit group
The bpstudel command is used to delete storage units or storage unit groups
# bpsturep -label <storage_unit_name> -cj 10 -maxmpx 2 -odo 0 : change the concurrent jobs ( number of drives available to use in the storage unit ) to 10, the maximum mpx per drive to 2 and turn off the on demand only flag
# bpsturep -label <storage _unit_group_name> -sm 2 : modify the selection method in a storage unit group to least recently used
The bpsturep command is used to modify a storage unit or storage unit group settings
# bpstudel -label <storage_unit_group_name>
# bpstudel -label <storage_unit_group_name>