Quote from Davina's email - "(2) Write the data to EDL with your desired
(long) browse time, clone it
to tape, then use a script to delete the save set on EDL when it reaches
the desired retention time, after checking that it has been successfully
cloned."
I support a site using EDL and have gone with option 2.
Networker Server is Solaris sparc 7.3.3 ,(possibly going to 7.4.3, another
subject), storage node the same.
Each data centre has a 60TB EDL and SL600 with 6 x LTO4 in each location.
Data is about 90-120TB a month unevenly split over the two and growing
rapidly.
Data is divided into pools based on browse/retention policy (anything from 1
month to 1 year), monthly cloning requirement, and life on the EDL.
Daily cloning uses a cloning utility (commercial, that I am involved with)
with wrapper scripts to keep X tape drives (X changes from day to day based
on monthly cloning and recovery needs) writing clones based on priority of
pools and data totals. Where possible I use the embedded storage node.
That's why I want to go to 7.4.3 for the -j switch on nsrclone. The utility
reports what's not cloned each day as well.
Monthly cloning runs through the month on X tape drives (first full/2 of the
month, not all data is retained monthly).
When free space on the EDL is low I a have a script that queries volumes by
pool/ saveset savetime/ age/ hostname / saveset name, then confirms all data
on the volumes are cloned, label the volumes to a scratch pool and recycle
them.
I'm currently refining it to be able to list X virtual volumes by working
its way through pools based on default rules and offer me volumes for
labelling.
I also need to add the ability to move data(clone) on the EDL so it stays on
the EDL as desired for quick recovery, but releases space on volumes, thus
avoiding excessive numbers of pools and the requirements for high device
counts during backup....... and so on.
Regards
Kit Cunningham
-----Original Message-----
From: Davina Treiber [mailto:Davina.Treiber AT PEEVRO.CO DOT UK]
Sent: Thursday, 6 November 2008 5:18 AM
Subject: Re: Changing the browse policy on a clone in 7.4
Esson, Paul wrote:
> Our set-up is Networker 7.4SP2 on AIX 5.3. We have commissioned an EMC
> Disk Library (DL4602) and also have a Quantum Scalar i2000 tape library.
> The intention is to write save sets to the DL in the first instance then
> clone them to the i2K. In 7.4 I know we can change the retention policy
> on the clone to be different from that of the original save set.
> However, how do we change the browse policy for the clone so that we
>
> do not have to do a save set recovery for these longer retention
> backups? Can we set a browse policy on the original that is greater
> than the retention policy to propagate this to the clone?
This is a problem for those environments where backups are written to an
EDL and later moved to tape. I am trying to do the same thing at the
moment for a couple of customers and working out the options. The
new(ish) functionality of having different retention policies for the
same save set in different pools is wasted because of the issue with the
browse time having to be not later than retention.
As I see it there are a few choices:
(1) Write the save sets to the EDL with your desired (long) browse time,
and use nsrstage rather than nsrclone to copy them for tape. There are
two disadvantages to this approach. The first is that you don't get the
data copied to tape until it has been on the EDL for a few days or
weeks. The second is that when you need the EDL to release space for new
backups it is probably busy staging your older data.
(2) Write the data to EDL with your desired (long) browse time, clone it
to tape, then use a script to delete the save set on EDL when it reaches
the desired retention time, after checking that it has been successfully
cloned.
(3) Write to EDL with a short retention time, clone to a pool with a
longer retention, then do some jiggery pokery with nsrck -L7 to
repopulate the index. Not a pretty solution.
None of these solutions are very good, but I think I'm going to have to
go with option 2 for the ones I am doing. Really we need EMC Engineering
to come up with a better solution. Perhaps we need a relaxation of the
rule where browse cannot be later than retention. Perhaps the best way
would be if browse could be later than retention for any particular
clone, but not longer than the longest retention - that would work.
To sign off this list, send email to listserv AT listserv.temple DOT edu and
type
"signoff networker" in the body of the email. Please write to
networker-request AT listserv.temple DOT edu if you have any problems with this
list. You can access the archives at
http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
To sign off this list, send email to listserv AT listserv.temple DOT edu and
type "signoff networker" in the body of the email. Please write to
networker-request AT listserv.temple DOT edu if you have any problems with this
list. You can access the archives at
http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
|
