Backup Central

Just wanted to see how people have their schedules configured for
various types of backups (system states and drive, exchange, sql, etc.).
In particular, I was wondering how people use Backup levels (i.e.
F,I,I,I,I,I,I or F,1,1,1,1,1,1 or F,1,2,3,4,5,6 etc) in their schedules.


I currently run a lot of my groups with F,I,I,F,I,I,I and was thinking
about removing the second full during the week but I am a little
paranoid about having a bad level FULL tape.

Any input would be helpful.

Cheers,
Jim

Note: To sign off this list, send a "signoff networker" command via email
to listserv < at > listserv.temple.edu or visit the list's Web site at
http://listserv.temple.edu/archives/networker.html where you can
should be sent to stan < at > temple.edu

On Tue, Apr 12, 2005 at 04:10:27PM -0400, Neild, Jim wrote:
Just wanted to see how people have their schedules configured for
various types of backups (system states and drive, exchange, sql, etc.).
In particular, I was wondering how people use Backup levels (i.e.
F,I,I,I,I,I,I or F,1,1,1,1,1,1 or F,1,2,3,4,5,6 etc) in their schedules.


I currently run a lot of my groups with F,I,I,F,I,I,I and was thinking
about removing the second full during the week but I am a little
paranoid about having a bad level FULL tape.

Any input would be helpful.

We found that most of our full backups were cold, unused data so we
don't run them as often. It might open us up to some data loss on a bad
full tape, but we can probably get most of that data back through
incremental recovers or older fulls. (But we also just jumped up from
DLT7 to LTO3, and with that much data on a tape we might rethink these
policies.)

We do a full every quarter, a 3 every month, a 5 every weekend, and an
incremental daily.

Dave

Note: To sign off this list, send a "signoff networker" command via email
to listserv < at > listserv.temple.edu or visit the list's Web site at
http://listserv.temple.edu/archives/networker.html where you can
should be sent to stan < at > temple.edu

If you are worried about minimizing dependency on your (less frequent)
fulls, another option (level) to consider is consolidation. It does a new
level 1, then combines that with the full that the level 1 is based on, to
create a new full. It's primarily intended for backing up remote clients on
expensive/slow WAN links, but I've also seen it used for large local clients
with mostly static data. Warning - It's slow and resource intensive on the
NetWorker server/library, has several other limitations and drawbacks. Do
your research before trying it. That being said, consolidation is something
to consider if it fits the situation...

CurtisE

-----Original Message-----
From: Legato NetWorker discussion
[mailto:NETWORKER < at > LISTSERV.TEMPLE.EDU]On Behalf Of Dave Mussulman
Sent: Tuesday, April 12, 2005 3:37 PM
To: NETWORKER < at > LISTSERV.TEMPLE.EDU
Subject: Re: [Networker] Schedules


On Tue, Apr 12, 2005 at 04:10:27PM -0400, Neild, Jim wrote:
Just wanted to see how people have their schedules configured for
various types of backups (system states and drive, exchange, sql, etc.).
In particular, I was wondering how people use Backup levels (i.e.
F,I,I,I,I,I,I or F,1,1,1,1,1,1 or F,1,2,3,4,5,6 etc) in their schedules.


I currently run a lot of my groups with F,I,I,F,I,I,I and was thinking
about removing the second full during the week but I am a little
paranoid about having a bad level FULL tape.

Any input would be helpful.

We found that most of our full backups were cold, unused data so we
don't run them as often. It might open us up to some data loss on a bad
full tape, but we can probably get most of that data back through
incremental recovers or older fulls. (But we also just jumped up from
DLT7 to LTO3, and with that much data on a tape we might rethink these
policies.)

We do a full every quarter, a 3 every month, a 5 every weekend, and an
incremental daily.

Dave

Note: To sign off this list, send a "signoff networker" command via email
to listserv < at > listserv.temple.edu or visit the list's Web site at
http://listserv.temple.edu/archives/networker.html where you can
should be sent to stan < at > temple.edu

Note: To sign off this list, send a "signoff networker" command via email
to listserv < at > listserv.temple.edu or visit the list's Web site at
http://listserv.temple.edu/archives/networker.html where you can
should be sent to stan < at > temple.edu

Neild, Jim wrote:
Just wanted to see how people have their schedules configured for
various types of backups (system states and drive, exchange, sql, etc.).
In particular, I was wondering how people use Backup levels (i.e.
F,I,I,I,I,I,I or F,1,1,1,1,1,1 or F,1,2,3,4,5,6 etc) in their schedules.


I currently run a lot of my groups with F,I,I,F,I,I,I and was thinking
about removing the second full during the week but I am a little
paranoid about having a bad level FULL tape.


I've worked on a lot of NetWorker sites, and almost all of them do a
weekly full with incrementals or level backups the other days. In
general if they do level backups they will do the same level each day
(e.g. L5) so that they can always recover from one full and one level
backup. Depending on the company some choose not to backup at weekends
if they are not 24x7 operations. Some choose to do their full backups
during the extended weekend window and some choose to spread the full
backups through the week.

I've never worked anywhere where the fulls are less frequent than
weekly. It amazes me to hear of people doing their fulls only monthly or
quarterly. These people must need to be paranoid about re-running any
failures ASAP.

Note: To sign off this list, send a "signoff networker" command via email
to listserv < at > listserv.temple.edu or visit the list's Web site at
http://listserv.temple.edu/archives/networker.html where you can
should be sent to stan < at > temple.edu

"if they do level backups they will do the same level each day
(e.g. L5) so that they can always recover from one full and one level
backup."

That's not quite right. If they do the same level each day (except the day
they do the full), they will need the media for the full and ALL level
backups since the full (for a complete restore).

In order to restore from the last full and (at most) 1 level, they would
need to decrement the level each day after the full (i.e. Full,9,8,7,6,5,4).

CurtisE

-----Original Message-----
From: Legato NetWorker discussion
[mailto:NETWORKER < at > LISTSERV.TEMPLE.EDU]On Behalf Of Davina Treiber
Sent: Wednesday, April 13, 2005 2:02 AM
To: NETWORKER < at > LISTSERV.TEMPLE.EDU
Subject: Re: [Networker] Schedules


Neild, Jim wrote:
Just wanted to see how people have their schedules configured for
various types of backups (system states and drive, exchange, sql, etc.).
In particular, I was wondering how people use Backup levels (i.e.
F,I,I,I,I,I,I or F,1,1,1,1,1,1 or F,1,2,3,4,5,6 etc) in their schedules.


I currently run a lot of my groups with F,I,I,F,I,I,I and was thinking
about removing the second full during the week but I am a little
paranoid about having a bad level FULL tape.


I've worked on a lot of NetWorker sites, and almost all of them do a
weekly full with incrementals or level backups the other days. In
general if they do level backups they will do the same level each day
(e.g. L5) so that they can always recover from one full and one level
backup. Depending on the company some choose not to backup at weekends
if they are not 24x7 operations. Some choose to do their full backups
during the extended weekend window and some choose to spread the full
backups through the week.

I've never worked anywhere where the fulls are less frequent than
weekly. It amazes me to hear of people doing their fulls only monthly or
quarterly. These people must need to be paranoid about re-running any
failures ASAP.

Note: To sign off this list, send a "signoff networker" command via email
to listserv < at > listserv.temple.edu or visit the list's Web site at
http://listserv.temple.edu/archives/networker.html where you can
should be sent to stan < at > temple.edu

Note: To sign off this list, send a "signoff networker" command via email
to listserv < at > listserv.temple.edu or visit the list's Web site at
http://listserv.temple.edu/archives/networker.html where you can
should be sent to stan < at > temple.edu

I've never worked anywhere where the fulls are less frequent than
weekly. It amazes me to hear of people doing their fulls only monthly or
quarterly. These people must need to be paranoid about re-running any
failures ASAP.

Hum! We do monthly full and I have no problem to sleep. But for doing this
and make sure everything is perfect, we clone all the full and weekly
levels. That way, we are sure there something on the original tape and it
is readable. So if a problem occur on a tape we always have an other copy
of it. I like it that way, I've seen to many place were they are thinking
they do backup but at the end, there were nothing useful or readable on
those tape.

Note: To sign off this list, send a "signoff networker" command via email
to listserv < at > listserv.temple.edu or visit the list's Web site at
http://listserv.temple.edu/archives/networker.html where you can
should be sent to stan < at > temple.edu

CurtisE wrote:
"if they do level backups they will do the same level each day
(e.g. L5) so that they can always recover from one full and one level
backup."


That's not quite right. If they do the same level each day (except the day
they do the full), they will need the media for the full and ALL level
backups since the full (for a complete restore).

In order to restore from the last full and (at most) 1 level, they would
need to decrement the level each day after the full (i.e. Full,9,8,7,6,5,4).


No you're incorrect. To quote from the Admin guide, a level backup
"backs up files that have changed since the last backup with a lower
numbered backup level". So if you always use full and level 5, the level
5 will back up all files that have changed since the full backup,
defined as a level 0. You would not need any other level 5 backups taken
in the mean time. What you are describing is an incremental backup.

Note: To sign off this list, send a "signoff networker" command via email
to listserv < at > listserv.temple.edu or visit the list's Web site at
http://listserv.temple.edu/archives/networker.html where you can
should be sent to stan < at > temple.edu

Alain Richard wrote:
I've never worked anywhere where the fulls are less frequent than
weekly. It amazes me to hear of people doing their fulls only monthly or
quarterly. These people must need to be paranoid about re-running any
failures ASAP.


Hum! We do monthly full and I have no problem to sleep. But for doing this
and make sure everything is perfect, we clone all the full and weekly
levels. That way, we are sure there something on the original tape and it
is readable. So if a problem occur on a tape we always have an other copy
of it. I like it that way, I've seen to many place were they are thinking
they do backup but at the end, there were nothing useful or readable on
those tape.

I suppose it's not too bad if you put some low level backups in on a
weekly basis, with higher levels or incrementals daily. The real
nightmare scenario would be a monthly full and daily incrementals, and
the server that goes down the day before the next full. I would not wish
to run a restore that needed a full and 30 incrementals to provide a
point-in-time recovery.

Note: To sign off this list, send a "signoff networker" command via email
to listserv < at > listserv.temple.edu or visit the list's Web site at
http://listserv.temple.edu/archives/networker.html where you can
should be sent to stan < at > temple.edu

You're right. I stand corrected. I incorrectly recalled
that a level backup picked up anything since the last
'same or' lower level. I hate getting old...

Please accept my apologies.

CurtisE

On Thu, 14 Apr 2005 15:34:55 +0100
Davina Treiber <Treiber < at > hotpop.com> wrote:
CurtisE wrote:
"if they do level backups they will do the same level
each day
(e.g. L5) so that they can always recover from one full
and one level
backup."


That's not quite right. If they do the same level each
day (except the day
they do the full), they will need the media for the full
and ALL level
backups since the full (for a complete restore).

In order to restore from the last full and (at most) 1
level, they would
need to decrement the level each day after the full
(i.e. Full,9,8,7,6,5,4).


No you're incorrect. To quote from the Admin guide, a
level backup
"backs up files that have changed since the last backup
with a lower
numbered backup level". So if you always use full and
level 5, the level
5 will back up all files that have changed since the
full backup,
defined as a level 0. You would not need any other level
5 backups taken
in the mean time. What you are describing is an
incremental backup.

Note: To sign off this list, send a "signoff networker" command via email
to listserv < at > listserv.temple.edu or visit the list's Web site at
http://listserv.temple.edu/archives/networker.html where you can
should be sent to stan < at > temple.edu

On Thu, Apr 14, 2005 at 03:39:48PM +0100, Davina Treiber wrote:
I suppose it's not too bad if you put some low level backups in on a
weekly basis, with higher levels or incrementals daily. The real
nightmare scenario would be a monthly full and daily incrementals, and
the server that goes down the day before the next full. I would not wish
to run a restore that needed a full and 30 incrementals to provide a
point-in-time recovery.

We were worried about that too... Doing a quarterly full and having to
do a full restore from the day before the next full. However, because
of our intermediate levels it's not that bad. The monthlies are a level
3, and the weeklies are a level 5. So, to do a full backup in the worst
case it's the full tapes, the last level 3, the last level 5, and all
the daily incrementals since that weekend (no more than 6.)

However, as you said in your previous message, making sure a machine
that misses a level 3 or level 5 doesn't keep getting incrementals is
very important. Otherwise, the dependency chain gets messed up and
media doesn't recycle. Some of the larger day-to-day time I spend with
Networker is tracking failures and moving them in/out of the rotation.

This was all with DLT. With LTO3, and new policies for shorter
retention times, I bet we'll adjust this. Even so, I can't imagine the
media involved for us to do a full backup of all of our systems every
week or two.

Dave

Note: To sign off this list, send a "signoff networker" command via email
to listserv < at > listserv.temple.edu or visit the list's Web site at
http://listserv.temple.edu/archives/networker.html where you can
should be sent to stan < at > temple.edu