I have small environment Master 9.1 + Appliance 4.1 (With connected Tape library having 4 FC LTO7 Tapes drives)
Appliance having FC connectivity but may be its not activated.
Tape Library 4 drive connected on Fiber.
Currently Tape backup writing data over the LAN
Storage Unit details:
Storage Device = HCART
DENSITY : HCART = 1/2 Inch Cartridge
Maximum concurrent Writen Drives: 4
Enabled Multiplexing Maximum Streams per drive: 2
Reduce fragment Size to : 1048575 Megabytes
I have attached appliance Fiber setting details.
Can you please suggest How I will improve my tape backup performance?
What is best value on Media Multiplexing and Is will complete my pending SLP tape backup job in case I will increase value from 2 to 20?
Not a appliance expert but I would ensure Netbackup has the following setting set:
NUMBER_DATA_BUFFERS = 256 or 512 (not default value)
SIZE_DATA_BUFFERS = 262144 (default value)
You can afterwards check the performance using the GEN_DATA_DIRECTIVE. You should get native LTO7 writing/reading speed.
See these tech note for reference:
NET_BUFFER_SZ, SIZE_DATA_BUFFERS and NUMBER_DATA_BUFFERS on UNIX - how they work and how to configure them. (Please disregard NET_BUFFER_SZ - setting is deprecated).
How to use the GEN_DATA file list directives with NetBackup for UNIX/Linux Clients for Performance Tuning
MPX = 4 seems to fair value, but increasing MPX count will not help on SLP backlog.
To expand on what @Nicolai said, media multiplexing can only help with direct backup to tape. Any type of duplication operation (SLP or manual) is only single streamed.
One thing that may help if this is not done already is to give each tape drive it's own FC port to operate from (and looking at the fibre settings this does look ot be the case).
Thanks Nicolai & David for reply.
Current Backup configuration:
Configured SLP duplication
1st Backup (Primary) goes to Appliance1
2nd backup goes form Appliance1 to Appliance2 (Over WAN)
3rd backup goes from Appliance2 to Tape Library (Over loacl LAN/FC)
I am still not cleared the "How to use the GEN_DATA file list directives with NetBackup" I will check again.
Below are the Appliance2 details:
NET_BUFFER_SZ, SIZE_DATA_BUFFERS and NUMBER_DATA_BUFFERS current configuration details as below :
# cat /usr/openv/netbackup/db/config/SIZE_DATA_BUFFERS = 262144
# cat /usr/openv/netbackup/db/config/NUMBER_DATA_BUFFERS = 30
# cat /usr/openv/netbackup/NET_BUFFER_SZ = 0
CLISH>Setting>FibreTransport Initiator Show
fc transport disabled
CLISH>Setting>FibreTransport Target Show
- [Info] Fibre Transport Deduplication is disabled.
- [Info] The Fibre Transport server is disabled.
Appliance2 Web: Setting-->Network-->Fibre Transport : find the attached snapshot for more information.
As per Web portal FC connectivity showing disabled.
But When I click it showing connected (Right now I am not save this setting)
Do have idea If I enable Fiber Transport target mode, What will happened?
Current success Tape backup job details (Appliance2 to Tape) : Showing CR received over FC, so confused that if attached file showing FC disabled, than how job details showing backup success over the FC?
Oct 20, 2021 12:38:26 PM - Info bptm (pid=27519) waited for full buffer 22241 times, delayed 58158 times
Oct 20, 2021 12:38:26 PM - end reading; read time: 0:03:45
Oct 20, 2021 12:38:31 PM - Info bpdm (pid=27605) completed reading backup image
Oct 20, 2021 12:38:31 PM - Info bpdm (pid=27605) EXITING with status 0
Oct 20, 2021 12:38:32 PM - Info appliance2 (pid=27605) StorageServer=PureDisk:appliance2; Report=PDDO Stats for (appliance2): read: 81936213 KB, CR received: 49485386 KB, CR received over FC: 0 KB, dedup: 0.0%
Oct 20, 2021 12:38:36 PM - Info bptm (pid=27519) EXITING with status 0 <----------
Oct 20, 2021 12:38:36 PM - end Duplicate; elapsed time 0:18:15
The requested operation was successfully completed. (0)
Tape drive leaning : No issue, I am doing regulars.
I assume that once I will enable Fiber Transport target mode tape backup will faster...
Change NUMBER_DATA_BUFFERS = 256 on all appliances
You don't have to restart any services. Once a new job starts, it will use the new setting. Jobs in the activity monitor will show buffers allocated (either 30 or the new value of 256).
You should see a big increase in tape drive performance with the new setting.
Create a policy, in the file list (what to be backed up, past below)
Ensure the retention is short - no reason to store random data for weeks or months.
As client - select the media server / appliance. The GEN_DATA directive does not work on Windows clients, it must be a UNIX/Linux client.
It is also good to know restoring data generated with GEN_DATA is not actual restored. Data is read, checked and thrown away.
Fibre transport mode is a way to perform backups (or duplications) over FC rather than network (over WAN). It modifies one (or more) of the FC ports to act as a target device rather than the default initiator mode. (in FC terms there needs to be initiators and targets - the server HBA's are normally initiators and devices like disk drives and tapes are targets).
This mode allows (SAN configured) clients to backup over FC rather than the network, or appliances to duplicate over FC rather than network.
Setting this mode will not make any difference in the way your tape backups/duplications occur other than reducing the available port count to use for tape drives.
The message saying no CR received over FC, is merely telling you that the data duplicated to tape was pulled from the local appliance and sent to tape (and not recieved by the appliance over FC). Of course the final path to tape is over FC.