SDLT drives replace with LTO2 with no increase in performance

[Graeme_Welsh]

Runnin Backup Exec 9.1, 5 servers, SSO options etc, using an HP MSL5000 library, have swopped out the SDLT drives with LTO2 and theres been no increase in performance
ive toyed with the idea of increasing the buffer count, although ive read that Symantec dont recommend changing buffer settings on Backup Exec
Using the veritas device drivers (version 5.1)

[STeve_O]

Have you tired OEM tape drivers?
Have you looked the upgrading firmware on the HBA and the switch?

[Graeme_Welsh]

not tried the OEM drivers,
when i installed the new drives, upgraded firmware on the library, drives and NSR
have now turned off 'write single block mode', and set the high water level to 7 under drive properties, i suspect these settings are there to give an illusion of control over backup exec
if i still see no improvement tonight, then will attempt to use the HP drivers..
find it quite shocking that theres been no improvement whatsoever

[Ken_Putnam]

Just for grins,

What are the Block Size, Buffer Size, and Number of Buffers

[Graeme_Welsh]

blcok size 64k
buffer size 64k
buffer count 10
high water 7

its rather annoying, on netbackup (another site), using the same library/drive config, and have the buffer count turned up to 128, using 64k buffer size, works a treat

[Ken_Putnam]

The numbers look OK. I might go to buffer of 128, but 10 64K buffers should work

When a job runs does the drive just hum? or can you heard it shoeshining?

And you did say that the single block and pass thru were all disabled, right?

[Graeme_Welsh]

the servers are in a remote site, so i cant really hear the drives, if i check in the web console of the library, it appears that the drives go idle/writing/idle etc when theyre running backup operations
im curious if there are any underlying SAN issues which might be preventing data being chucked at the drives consistently, however, no errors are reported on any of the fibre switches or the NSR of the library
pass through wasnt disabled, it is now tho

[Ken_Putnam]

it appears that the drives go idle/writing/idle


This would indeed cause a reduction. possibly drastic in throughput

[Graeme_Welsh]

yes, indeed,
question is why tho
HP drivers made no difference, exchange database backups are slightly quicker, will try and defrag the drives, see if that helps

[Ken_Putnam]

If the data is on a different server, try just copying a large amount of data to the media server. if this has high transfer rate you can at least get rid of the network as a cause of the slow writes to tape

[Guest_User]

How fast is the HBA that the MSL5000 is connected to? LTO2 requires a minimum of Ultra160, and Ultra320 is preferred.

[Gauri_Ketkar]

Hi

Could you please update us on the issue ..


Revert for any further Query ..



Thank you
Gauri