Slow in Backup

[will68]

Hi,

My using HP 1/8 G2 LTO-4 Ultr 1760 autoloader and LTO4 Ultrium 1.6TB tape. I realise my backup took more than 4 days to back 1.5TB of data. This does not tally with the backup transfer rate that can be supported by the loader. Any idea where could went wrong? My data is mounted as a shared drive to my backup server which is connected to the tape library.

[Ralf Loehmann_2]

Hi,

you might be just to slow, because the LTO drives have a minimum performance. If you get slower than this minimum speed the drive has to wait for new data and rewind than to start writing again. The drives are writing on the fly and that is the reason why below that speed you loose performance nearly exponential. Also you might keep in mind that this start/stops will reduce tape media and head life.

Here is a document, which explains this and might help to improve your performance:
http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=lpg50460#A2

[Trygve Henriksen]

'Mounted as a shared Drive'?

Do you mean that you've connected up a disk from another server as a networked drive, or that it's a local disk(on the backup server) which the other servers/users/clients connect to to store data?

A drive on another server is to ask for trouble...

How is the tape drive connected to your server?
(Fibre channel, eSata, SCSI of some sort?)
Is it using the same controller as the disks?

Which server do you have?

We had an older Proliant DL server(ancient, standard PCI, about 800MHz) which we managed to plug a FC card into to hook up to our first SAN(MSL6030 with twin LTO3, MSA1000).
When it backed up local drives, it barely managed to push enough data to the drive, but as soon as it started accessing the SAN drive, it faltered because the PCI-bus couldn't support more than the lowest throughput the FC-card required. And backing up the SAN drive meant reading data from the san, through the CD, then spitting it out through the FC again, towards the Robot. With the OS multitasking it tried to do both, on the same controller, without having the bandwidth.
(We moved the SAN drive to another server, and have later put the server out of its misery entirely)