Even in times where everyone is talking about cloud storage, tape drives are still important storage devices. On the one hand, the magnetic tapes used by them are pretty inexpensive (price per terabyte), on the other hand, tape drives are very robust and reliable devices.
Nevertheless, tape drives also have some disadvantages: since tapes can only be read and written sequentially (at least when using Backup Exec) you can only write one data stream at a time to them.
If a tape drive is in use by a job in Backup Exec and another job starts that requires the same drive, it is queued until the first job finishes and the device is released.
In addition to this, magnetic tapes need quite a long positioning time to find a dedicated file on them.
While a file written to a hard disk can be approached within milliseconds, it may take multiple minutes of fast forwarding a tape to find the same file.
Another disadvantage of tape drives is that they can only write with a defined speed.
If the data stream is interrupted during a write operation, the tape drive has to stop, reposition the tape and wait for the buffers to fill up again before it can continue writing to the tape again. This process needs quite some time which will increase the overall backup time window, if it happens too often. Therefore it is important to design the whole backup environment to be able to keep the tape drive(s) in streaming mode.
In many environments it’s better to use a tape library with two LTO 5 drives than one with a single LTO 6 drive, because the LTO 6 drive will not run in streaming mode, because the environment cannot deliver the data fast enough.
Of course, there is the opportunity to use a backup-to-disk storage locally attached to the backup server in order to cache the data before it is written to tape, but I only recommend this, if the disk-based backup is seen as the primary backup and tape is just an offsite copy.