Tape Library Deployment
Prerequisites for Tape Library Deployment
A physical node in the Enterprise is available to control the tape library’s media changer. This can be a Windows server, Linux server, or the NetApp storage controller. A virtualized node cannot be used for media changer control. This controller node must have a host bus adapter. Note that a media changer is also referred to as a robotic arm and a host bus adapter is also referred to as an HBA.
Recommended Best Practices for Tape Libraries
The media changer is attached to a SAN (storage area network) and zoned to the master server. Therefore the master server acts as the tape library’s controller node.
The tape drives are attached to the SAN and zoned to the NetApp storage system. Therefore the NetApp storage system acts as the device server.
The tape library is a modern, vendor-supported library with tape drives capable of writing in variable block size mode. The media changer is SMC-2 compliant and supports the DVCID option of the Read Element Status command. The tape drives are SCSI-2 compliant and support variable length records and extended file marks. The tape library and tapes support bar coding. Most modern tape libraries fit these requirements. DPX does not support libraries and tape drives currently unsupported by the device vendor.
For DPX purposes, the media changer device and associated tape drives must belong to only one Enterprise controlled by a single master server. Tape library media changer control cannot be shared with other backup products and cannot be accessed by more than one master server.
Other Tape Library Setup Scenarios
If, in conjunction with the customer, a strategic determination is made to attach the media changer or tape drives differently, you can do so. However, the practices described in Recommended Best Practices for Tape Libraries are highly recommended to maximize reliability and minimize troubleshooting challenges.
There are several supported ways to attach the tape library’s media changer and tape drives to the Enterprise. They can be connected through a fiber channel SAN, iSCSI SAN, direct fiber channel attachment, or legacy parallel SCSI cable attachment.
If the connection is through a fiber channel SAN, the media changer must be controlled by a device server, not by a NetApp storage system, and the tape drives can be controlled by any node or NetApp storage system.
If direct attachment through a fiber channel or legacy parallel SCSI is required, it is highly recommended for the media changer and at least one tape drive to be attached directly to the master server to support media changer control and Catalog backups. If a tape drive cannot be connected or accessed through the master server, an alternate strategy must be used to move Catalog backups to a safe medium that can be recovered in the event of a site disaster.
iSCSI SAN connections are generally not recommended due to latency and connection reliability. If iSCSI is absolutely necessary, for example, if no physical server with HBA is available for media changer control, great care must be taken by the site administrators to ensure such connections are very reliable. Any disruption in connection will impact backup and media control operations.
Though it is possible to attach a tape library directly to a NetApp storage system for exclusive control of media changer and all tape devices, it is not recommended as an alternate strategy for Catalog backups is required. Additionally, media changer reliability and troubleshooting may be problematic, requiring moving the media changer device to a node with an HBA physical server. NetApp media changer control is also limited to tape libraries with 600 total slots or less; larger tape libraries do not work properly due to SCSI limitations.
Last updated