NBU 6.5 Settings - RFC

[ADDODD]

hey gang,

I'm in the process of rolling out a new NBU 6.5(.4) environment here.

I was hoping to get some feedback on settings and so on as I'm not an NBU Guru.  I've been going through the Admin Guide and the online help and the forums looking for information on "best practices"... I'm hoping these will serve as a good / reasonable starting point.

If any of these look odd or out of place, can you please shout and let me know why?  I know a lot of this is a "black art" and depends on the specifics of the environment (servers, tape drives, LAN, etc)... The environment is pretty big (2x Solaris Master Servers clustered, 2x Solaris Media servers, 8 LTO-4 tape drives in a storage tek library, GigE LAN).

Here's what I'm planning to use:

in Master Server, Properties, Global Attributes:
----
Job Retry Delay: 15 minutes
Max Jobs per Client: 2
Schedule Backup Attempts: 2 tries / 12 hours
Maximum Backup copies: 4
Policy Update Interval: 10 min
Compress Catalog Interval: 16 days
Priority of restore jobs: 99999
Maximum Vault Jobs: 50

--------------------------------------------------------------------------------
In Master Server, Properties, Clean-up:
---
Keep Logs for: 28 days
Keep True Image Restore for: 1 day
Keep Vault Logs for: 30 days
Move restore job from incomplete to done after 1 day
move backup job from incomplete to done after 4 hours

-----------------------------------------------------------------------------------------------------
Storage Unit Configuration:
---
STU 1 (Tape Drives for Weekly Full backups, Monthly Full Backups, Annual Full Backups and vaulting):
Type: Media Manager
Device: robot (ACS)
Media Server: (Multiple, shared drives with SSO)
Maximum Concurrent Write Drives: 8
Enable Multiplexing: checked
Maximum data streams: <<NOT SURE, HOW DO YOU DECIDE?>>
Reduce Fragment Size to: 2048 MB
---
STU 2 (Daily Incrementals):
Type: Disk, Basic Disk
Media Server: multiple
Path: /path/to/storage
Maximum Concurrent jobs: 4 (this is pretty fast SAN disk on multiport HBAs.  can it go higher?)
High water mark: 90
Low water mark: 85
Reduce Fragment Size: 2048

----------------------------------------------------------------------------------------------------------
Policy (General Configuration):
---
Take Checkpoints every: 10 minutes
Limit jobs per policy: 8
Allow Multiple data streams: <checked>
*Schedules, Attributes, Multiple copies: 2
-Primary copy to tape media pool
-Secondary copy to Offsite media pool
*Schedule, Attributes, Media Multiplexing: 4

Comments (good and bad) welcome!

thanks,

ADDODD

[Amit_Karia]

Keep Logs for: 28 days : Wanted to know the reason for these many days.. it may fill up the catalog hence your emm database may not start.. suggest you should keep it for 1 week or even lesser depending on your catalog size

Keep Vault Logs for: 30 days : Again you should not keep it for more than a week (in case if you dont want to refer to vaulted images)

Maximum data streams: <<NOT SURE, HOW DO YOU DECIDE?>> maximum streams depends  on the number of filesystems or partitions you have to backup at the same time

Media Server: multiple : Is the same volume shared between two media servers? or a single media server

[ADDODD]

Keep logs for 28 days: i thought this referred to the log files, not the catalog.   I had "Compress Catalog Interval: 16 days" set in the General Attributes... whereas the "keep logs for X days" referred to /opt/openv/netbackup/logs/* debug logs and stuff (or maybe to /opt/openv/logs for the unified logging?)?

Keep Vault Logs for 30 days: I thought I'd need 30 days in order for things like vault reports to work.  if that's not the case, i'll set them down to a week or two.

Maximum data streams: ah, ok, so a high number will probably be ok then.  32 (higher?)... I'd like to have roughly 18-20 jobs per media server (x 2 media servers) running with multiplexing.

Media Server: Multiple: there's 2 media servers.  each will have it's own (4 TB) storage area for disk staging, or backup to disk.

more input welcome!

thanks,

ADDODD

[Nicolai]

You need to configure NUMBER_DATA_BUFFERS, NUMBER_DATA_BUFFERS_RESTORE and SIZE_DATA_BUFFERS on media servers.

Just to show what a difference the buffer settings really do for a LTO3 drive:

Before tuning: 19MB/sec (default values)
Second tuning attempt: 49MB/sec (using 128K block size, 64 buffers)
Final result: 129MB/sec (using 128K block size and 256 buffers)

NetBackup 6.5 Backup Planning and Performance Tuning Guide

Netbackup buffer tuning

[ADDODD]

hey,

I'm planning on doing some BUFFER_* tuning as well, I just haven't got there yet.  Thanks for the tips though!

As far as the other General / Global settings go... anything else anyone can/cares to point out?

thanks,

ADDODD