Veritas-bu

[Veritas-bu] Is it wrong...

2001-06-06 13:47:39
Subject: [Veritas-bu] Is it wrong...
From: scott.kendall AT abbott DOT com (scott.kendall AT abbott DOT com)
Date: Wed, 6 Jun 2001 12:47:39 -0500
If all of the below is true...

1.) Storage Units are NOT configured as on-demand
2.) number of drives configured on each Storage Unit matches the physical
number of drives
3.) the classes are using any_availalbe for their Storage Unit selection

... with the way NetBackup chooses Storage Units alphabetically, it will use
all the drives in the robot via the first Storage Unit and the second Storage
Unit will just sit waiting to be used, but can't because SSO is out of drives
in that library.

With that said, be careful of the following scenario... multiple media servers
sharing the same robot.  All Storage Units configured for ALL of the drives in
the library and not on-demand and classes configured to use any_available
Storage Unit.  What will happen is that the media server with the
alphabetically first Storage Unit will do most of the work while the other
media servers will just sit there.  Now, let's say that there are 12 drives in
the library and 3 media servers sharing the library.  In my opinion it would
be the wrong solution to take the first four drives and give them to server1
(and possibly share with other media servers using on-demand STU's to only
back themselves up over the SAN) the second four to server2 and the last four
to server3.  Server 1's drives would always be used first.  Any overflow when
there is a large number of jobs would go to server2's and then finally to
server3's.  A better way to ensure that all drives are being used equally and
that the load is shared by the media servers is to configure all three to use
all 12 of the drives.  Then configure the maximum number of drives on each
media server's Storage Unit to a number less than the number of physical
drives.  Server1's Storage Unit will still be used first, but this time will
randomly pick any one of the 12 drives to be used.  If the STU is configured
for a max of 4 drives (because we have determined that server1 can only pull
enough data over the network to push 4 drives, not necessarily because 4
happens to be 1/3 of the library and there are 3 media severs in this example)
when it is using all 4 drives, server2's STU will start to be used and then
eventually server3's.  In this scenario server3 is only used when we need the
extra bandwidth/horsepower, but all of the drives are being randomly used so
the wear across the entire library is even.

I'm not doing it, but it makes sense to be able to expand your functionality
with ideas like setting up two different storage units to the same drives on
the same robot from the same media server if you are using SSO in a SAN and
have the requirement.  Each Storage Unit could have different settings such as
max mpx or fragment size, even though they are sharing the exact same drives.
They could be configured as on-demand and called by the classes with that
particular requirement.


- Scott



                                                                                
                                                   
                    "Peter Smith"                                               
                                                   
                    <petes AT convergentdata DOT net>           To:     "'David 
A. Chapa'" <david AT xbpadm-commands DOT com>,                   
                    Sent by:                             <veritas-bu AT 
mailman.eng.auburn DOT edu>                                       
                    veritas-bu-admin AT mailman DOT eng.        cc:             
                                                          
                    auburn.edu                           Subject:     RE: 
[Veritas-bu] Is it wrong...                              
                                                                                
                                                   
                                                                                
                                                   
                    06/05/2001 11:54 AM                                         
                                                   
                    Please respond to petes                                     
                                                   
                                                                                
                                                   
                                                                                
                                                   




I doesn't a first glance look like a problem. But it is logically wrong.
There is no need to reference the robot twice. This would only make sense if
you were limiting the number of drives logically defined for a storage unit.

As far as being a potential problem, I would say probably no only as long as
for the classes accessing these storage units they don't try to utilize all
10 drives associated with one storage unit. Of course, NetBackup will
probably just queue any jobs trying to access the same drives.

Peter Smith
Convergent Data Services Inc.
www.convergentdata.net
720-482-9559 (work)
720-482-0907 (fax)


-----Original Message-----
From: veritas-bu-admin AT mailman.eng.auburn DOT edu
[mailto:veritas-bu-admin AT mailman.eng.auburn DOT edu]On Behalf Of David A.
Chapa
Sent: Tuesday, June 05, 2001 10:37 AM
To: veritas-bu AT mailman.eng.auburn DOT edu
Subject: [Veritas-bu] Is it wrong...


I have a client with the following configuration:

L700 - in a SAN with NetBackup 3.4 and SS0

Upon reviewing their site, I noticed several Storage
Units configured for each media server.  They each
contained the same number of drives and referenced the
same robot number (0).  I was concerned at first
glance, but then realized that functionally this wasn't
a problem.  However, could it eventually become a
problem?  Could contention be an issue down the road?

Example

MEDIASRV1
  STU1   Robot (0)  Number of Drives (10)
  STU2   Robot (0)  Number of Drives (10)

MEDIASRV2
  STU3   Robot (0)  Number of Drives (10)
  STU4   Robot (0)  Number of Drives (10)

Another brain on this one would help...

thanks

david

<><><><><><><><><><><><><><><><><><><><>
David A. Chapa
Consulting Manager
DataStaff, Inc.
847 413 1144
http://www.consulting.datastaff.com
---------------------------------------
http://www.xbpadm-commands.com
NBU-LSERV AT datastaff DOT com - Adv. Scripting
_______________________________________________
Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu

_______________________________________________
Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu





<Prev in Thread] Current Thread [Next in Thread>