cancel
Showing results for 
Search instead for 
Did you mean: 

backup duplication performance issues

Jan_Rustico
Level 4

I have a similar issues as discussed in http://www.veritas.com/community/forums/using-bpdu... .​  I want to clarify if we list and identify the images need to cancelled and cancelled it, shall we call the cancelled backup duplicatye images again to be process for tape duplication?

13 REPLIES 13

sdo
Moderator
Moderator
Partner    VIP    Certified

AFAIK, if you cancel the SLP processing from the CLI, or cancel SLP by image from the CLI, then all incomplete stages for the SLP or image are cancelled.

 

 

sdo
Moderator
Moderator
Partner    VIP    Certified

If you show us the output from these, then we might be able to help you spot a configuration problem:

bpstulist -L -g

nbstl -L

nbstlutil report

Jan_Rustico
Level 4

Please see the attached outcomes.

Nicolai
Moderator
Moderator
Partner    VIP   

If you cancel a backup image under SLP control, you have to duplicate it manual. It is not possible to re-SLP a backup image that has either been canceled or marked as complete.

Furthermore, if you cancel a backup image, it may expire immediately, depending on retention period.

A backup image under SLP control has infinity retention, when either SLP is being completed or canceled the original retention is applied.

 

 

Jan_Rustico
Level 4

Ok, since we are not able to re-run it again, how can we improve the backup duplication performance (tape library)? We dont want to lose any data here. Any suggestions?

Nicolai
Moderator
Moderator
Partner    VIP   

Ensure you have NUMBER_DATA_BUFFERS and SIZE_DATA_BUFFERS configured

in [INSTALL_PAHT]/Netbackup/db/config

create file NUMBER_DATA_BUFFERS and put 256 or 512 in it

create file SIZE_DATA_BUFFERS and put 262144 in it

file must be called exactly as above - not dot extension are allowed. That should speed up tape writes a lot.

http://www.veritas.com/docs/000016306

Jan_Rustico
Level 4

What is the recommended values for NUMBER_DATA_BUFFERS and SIZE_DATA_BUFFERS? I've attached my master/media server configuration/system information

Also here's the tape library/tape cartridge info:

LTO Ultrium 6 Tape Drive
2.5 TB native cartridge capacity (up to 6.25 TB with 2.5:1 data compression)
576 GB/hour (160 MBps) native data transfer rate
6 Gbps SAS connectivity

Nicolai
Moderator
Moderator
Partner    VIP   

NUMBER_DATA_BUFFERS = 256

SIZE_DATA_BUFFERS = 262144

Pretty universal settings providing good performance on all systems. Please see my previous post on how to configure or use the tech note I linked to.

Jan_Rustico
Level 4
How about the buffer size in the back-up client's registry? What is the recommended value for my current configuration?
 
HKEY_LOCAL_MACHINE\SOFTWARE\VERITAS\NetBackup\CurrentVersion\Config\
REG_DWORD: Buffer_Size: <Decimal value in KB>

Nicolai
Moderator
Moderator
Partner    VIP   

Leave client registry to default - 0

Historical NET_BUFFER_SZ on the client  used to be a discussion subject , but now has no meaning by now becuase of modern TCP/IP stack, NIC TCP offload etc etc.

Best practices for NET_BUFFER_SZ and Buffer_size, why can't NetBackup change the TCP send/receive space

http://www.veritas.com/docs/000026262

Jan_Rustico
Level 4

can we also modify the communications buffer size and raw partition read buffer size? refer to the attached image.

Nicolai
Moderator
Moderator
Partner    VIP   

Never heard the need to tune those, so use default values.

Mark_Solutions
Level 6
Partner Accredited Certified

You can modify the communications buffer if you wish but not usually needed.

The raw partition size buffer is usually done on the media server that handles clients doing Flashbackup policies and a setting of 1mb works really well for that .. but otherwise not needed

All setting discussed in the tuning guide which is well worth a read and should answer all of your tuning questions:

http://www.veritas.com/docs/000004178