Backup data format
 

Hi,
What formats are the full and incremental dumps from a NetApp OnTAP
system in? As in, is the backup data in some proprietary NetApp
format?

Also, what is the difference between NDMP and SnapMirror?

Dvy

Backup data format
 

On 5 Feb 2007 13:37:57 -0800, wrote:

Hi,
What formats are the full and incremental dumps from a NetApp OnTAP
system in? As in, is the backup data in some proprietary NetApp
format?

It's just plain 'ol *nix DUMP/RESTORE. Nothing special or
proprietary. It can even be restored from a Solaris machine (restore)
if you fsf past the NBU headers.


Also, what is the difference between NDMP and SnapMirror?

NDMP just uses dump/restore over an NDMP socket. NDMP is a command
protocol, not a data format. It merely handles the commands between
filer and application, the data format is still dump just like any
unix machine.

Snapmirror is not a data format either, it is a transport protocol.
It just happens to be transporting at a block level, the interface of
which is proprietary to WAFL.

The difference between them? NDMP is file level, snapmirror is block.
This can be a tremendous savings in bandwidth depending on your data
set.
Snapmirror is configurable on the filer, NDMP has to be managed
remotely (ie. cron).
Snapmirror has no limits to the number of incrementals it can run.
NDMP (because of dump's limitation) can only run 9 incremental levels.
There are some variations to this but it's generally true.
Snapmirror will delete data on the destination that is removed from
the source. NDMP will only append or add, not delete.

There may be more differences but those are a quick reference.

~F

Backup data format
 

On Feb 5, 6:32 pm, Faeandar wrote:
On 5 Feb 2007 13:37:57 -0800, wrote:

Hi,
What formats are the full and incremental dumps from a NetApp OnTAP
system in? As in, is the backup data in some proprietary NetApp
format?

It's just plain 'ol *nix DUMP/RESTORE. Nothing special or
proprietary. It can even be restored from a Solaris machine (restore)
if you fsf past the NBU headers.



Also, what is the difference between NDMP and SnapMirror?

NDMP just uses dump/restore over an NDMP socket. NDMP is a command
protocol, not a data format. It merely handles the commands between
filer and application, the data format is still dump just like any
unix machine.

Snapmirror is not a data format either, it is a transport protocol.
It just happens to be transporting at a block level, the interface of
which is proprietary to WAFL.

The difference between them? NDMP is file level, snapmirror is block.
This can be a tremendous savings in bandwidth depending on your data
set.
Snapmirror is configurable on the filer, NDMP has to be managed
remotely (ie. cron).
Snapmirror has no limits to the number of incrementals it can run.
NDMP (because of dump's limitation) can only run 9 incremental levels.
There are some variations to this but it's generally true.
Snapmirror will delete data on the destination that is removed from
the source. NDMP will only append or add, not delete.

There may be more differences but those are a quick reference.

~F

Thanks

When I have a tier 1 and tier 2 NAS environment (tier 1 is the newer
FAS equip and tier 2 run older hardware with slower drives), should I
run NDMP or SnapMirror to save my dumps on tier 2 for a while before
moving them to tape or is it better to go straight from tier 1 to
tape? That is, I've heard some people use tier 2 as a staging
environment for a while for the backups before they move them to
tape. I am just trying to understand best practices here...

Dvy

Backup data format
 

On 6 Feb 2007 09:55:47 -0800, wrote:

On Feb 5, 6:32 pm, Faeandar wrote:
On 5 Feb 2007 13:37:57 -0800, wrote:

Hi,
What formats are the full and incremental dumps from a NetApp OnTAP
system in? As in, is the backup data in some proprietary NetApp
format?

It's just plain 'ol *nix DUMP/RESTORE. Nothing special or
proprietary. It can even be restored from a Solaris machine (restore)
if you fsf past the NBU headers.



Also, what is the difference between NDMP and SnapMirror?

NDMP just uses dump/restore over an NDMP socket. NDMP is a command
protocol, not a data format. It merely handles the commands between
filer and application, the data format is still dump just like any
unix machine.

Snapmirror is not a data format either, it is a transport protocol.
It just happens to be transporting at a block level, the interface of
which is proprietary to WAFL.

The difference between them? NDMP is file level, snapmirror is block.
This can be a tremendous savings in bandwidth depending on your data
set.
Snapmirror is configurable on the filer, NDMP has to be managed
remotely (ie. cron).
Snapmirror has no limits to the number of incrementals it can run.
NDMP (because of dump's limitation) can only run 9 incremental levels.
There are some variations to this but it's generally true.
Snapmirror will delete data on the destination that is removed from
the source. NDMP will only append or add, not delete.

There may be more differences but those are a quick reference.

~F

Thanks

When I have a tier 1 and tier 2 NAS environment (tier 1 is the newer
FAS equip and tier 2 run older hardware with slower drives), should I
run NDMP or SnapMirror to save my dumps on tier 2 for a while before
moving them to tape or is it better to go straight from tier 1 to
tape? That is, I've heard some people use tier 2 as a staging
environment for a while for the backups before they move them to
tape. I am just trying to understand best practices here...

Dvy

Well, it truly depends on your budget, your sensitivity to performance
degradation and your retention needs.

People who stage usualy do one of two things:

1) Snapvault to secondary and NDMP backup to tape from there.
2) Use their secondary as a disk staging unit managed by their backup
application.

if you are performance sensitive and/or want long retentions, then
Snapvault is the way to go (assuming NetApp for both tiers).
The incremental block updates take less time and resources to transfer
and you can setup a very long retention schedule for snapshots on the
tier2 disk providing you have enough of it.

Best practices in cases like this are completely based on your
business needs.

~F

Backup data format
 

On Feb 6, 12:50 pm, Faeandar wrote:
On 6 Feb 2007 09:55:47 -0800, wrote:



On Feb 5, 6:32 pm, Faeandar wrote:
On 5 Feb 2007 13:37:57 -0800, wrote:

Hi,
What formats are the full and incremental dumps from a NetApp OnTAP
system in? As in, is the backup data in some proprietary NetApp
format?

It's just plain 'ol *nix DUMP/RESTORE. Nothing special or
proprietary. It can even be restored from a Solaris machine (restore)
if you fsf past the NBU headers.

Also, what is the difference between NDMP and SnapMirror?

NDMP just uses dump/restore over an NDMP socket. NDMP is a command
protocol, not a data format. It merely handles the commands between
filer and application, the data format is still dump just like any
unix machine.

Snapmirror is not a data format either, it is a transport protocol.
It just happens to be transporting at a block level, the interface of
which is proprietary to WAFL.

The difference between them? NDMP is file level, snapmirror is block.
This can be a tremendous savings in bandwidth depending on your data
set.
Snapmirror is configurable on the filer, NDMP has to be managed
remotely (ie. cron).
Snapmirror has no limits to the number of incrementals it can run.
NDMP (because of dump's limitation) can only run 9 incremental levels.
There are some variations to this but it's generally true.
Snapmirror will delete data on the destination that is removed from
the source. NDMP will only append or add, not delete.

There may be more differences but those are a quick reference.

~F

Thanks

When I have a tier 1 and tier 2 NAS environment (tier 1 is the newer
FAS equip and tier 2 run older hardware with slower drives), should I
run NDMP or SnapMirror to save my dumps on tier 2 for a while before
moving them to tape or is it better to go straight from tier 1 to
tape? That is, I've heard some people use tier 2 as a staging
environment for a while for the backups before they move them to
tape. I am just trying to understand best practices here...

Dvy

Well, it truly depends on your budget, your sensitivity to performance
degradation and your retention needs.

People who stage usualy do one of two things:

1) Snapvault to secondary and NDMP backup to tape from there.
2) Use their secondary as a disk staging unit managed by their backup
application.

if you are performance sensitive and/or want long retentions, then
Snapvault is the way to go (assuming NetApp for both tiers).
The incremental block updates take less time and resources to transfer
and you can setup a very long retention schedule for snapshots on the
tier2 disk providing you have enough of it.

Best practices in cases like this are completely based on your
business needs.

~F

Thanks... I take it then that it is generally a good practice to, in
some way or another, use the tier 2 as a stage prior to tape?
(assuming we have enough space on tier2)

Backup data format
 

On 6 Feb 2007 15:16:55 -0800, wrote:

On Feb 6, 12:50 pm, Faeandar wrote:
On 6 Feb 2007 09:55:47 -0800, wrote:



On Feb 5, 6:32 pm, Faeandar wrote:
On 5 Feb 2007 13:37:57 -0800, wrote:

Hi,
What formats are the full and incremental dumps from a NetApp OnTAP
system in? As in, is the backup data in some proprietary NetApp
format?

It's just plain 'ol *nix DUMP/RESTORE. Nothing special or
proprietary. It can even be restored from a Solaris machine (restore)
if you fsf past the NBU headers.

Also, what is the difference between NDMP and SnapMirror?

NDMP just uses dump/restore over an NDMP socket. NDMP is a command
protocol, not a data format. It merely handles the commands between
filer and application, the data format is still dump just like any
unix machine.

Snapmirror is not a data format either, it is a transport protocol.
It just happens to be transporting at a block level, the interface of
which is proprietary to WAFL.

The difference between them? NDMP is file level, snapmirror is block.
This can be a tremendous savings in bandwidth depending on your data
set.
Snapmirror is configurable on the filer, NDMP has to be managed
remotely (ie. cron).
Snapmirror has no limits to the number of incrementals it can run.
NDMP (because of dump's limitation) can only run 9 incremental levels.
There are some variations to this but it's generally true.
Snapmirror will delete data on the destination that is removed from
the source. NDMP will only append or add, not delete.

There may be more differences but those are a quick reference.

~F

Thanks

When I have a tier 1 and tier 2 NAS environment (tier 1 is the newer
FAS equip and tier 2 run older hardware with slower drives), should I
run NDMP or SnapMirror to save my dumps on tier 2 for a while before
moving them to tape or is it better to go straight from tier 1 to
tape? That is, I've heard some people use tier 2 as a staging
environment for a while for the backups before they move them to
tape. I am just trying to understand best practices here...

Dvy

Well, it truly depends on your budget, your sensitivity to performance
degradation and your retention needs.

People who stage usualy do one of two things:

1) Snapvault to secondary and NDMP backup to tape from there.
2) Use their secondary as a disk staging unit managed by their backup
application.

if you are performance sensitive and/or want long retentions, then
Snapvault is the way to go (assuming NetApp for both tiers).
The incremental block updates take less time and resources to transfer
and you can setup a very long retention schedule for snapshots on the
tier2 disk providing you have enough of it.

Best practices in cases like this are completely based on your
business needs.

~F

Thanks... I take it then that it is generally a good practice to, in
some way or another, use the tier 2 as a stage prior to tape?
(assuming we have enough space on tier2)


Not necessarily. tier2 is great for seldom used data, or archive
data, or whatever else you can think of such that it doesn't need to
be on expensive tier1 disk.

~F

Backup data format
 

On Feb 6, 4:15 pm, Faeandar wrote:
On 6 Feb 2007 15:16:55 -0800, wrote:



On Feb 6, 12:50 pm, Faeandar wrote:
On 6 Feb 2007 09:55:47 -0800, wrote:

On Feb 5, 6:32 pm, Faeandar wrote:
On 5 Feb 2007 13:37:57 -0800, wrote:

Hi,
What formats are the full and incremental dumps from a NetApp OnTAP
system in? As in, is the backup data in some proprietary NetApp
format?

It's just plain 'ol *nix DUMP/RESTORE. Nothing special or
proprietary. It can even be restored from a Solaris machine (restore)
if you fsf past the NBU headers.

Also, what is the difference between NDMP and SnapMirror?

NDMP just uses dump/restore over an NDMP socket. NDMP is a command
protocol, not a data format. It merely handles the commands between
filer and application, the data format is still dump just like any
unix machine.

Snapmirror is not a data format either, it is a transport protocol.
It just happens to be transporting at a block level, the interface of
which is proprietary to WAFL.

The difference between them? NDMP is file level, snapmirror is block.
This can be a tremendous savings in bandwidth depending on your data
set.
Snapmirror is configurable on the filer, NDMP has to be managed
remotely (ie. cron).
Snapmirror has no limits to the number of incrementals it can run.
NDMP (because of dump's limitation) can only run 9 incremental levels.
There are some variations to this but it's generally true.
Snapmirror will delete data on the destination that is removed from
the source. NDMP will only append or add, not delete.

There may be more differences but those are a quick reference.

~F

Thanks

When I have a tier 1 and tier 2 NAS environment (tier 1 is the newer
FAS equip and tier 2 run older hardware with slower drives), should I
run NDMP or SnapMirror to save my dumps on tier 2 for a while before
moving them to tape or is it better to go straight from tier 1 to
tape? That is, I've heard some people use tier 2 as a staging
environment for a while for the backups before they move them to
tape. I am just trying to understand best practices here...

Dvy

Well, it truly depends on your budget, your sensitivity to performance
degradation and your retention needs.

People who stage usualy do one of two things:

1) Snapvault to secondary and NDMP backup to tape from there.
2) Use their secondary as a disk staging unit managed by their backup
application.

if you are performance sensitive and/or want long retentions, then
Snapvault is the way to go (assuming NetApp for both tiers).
The incremental block updates take less time and resources to transfer
and you can setup a very long retention schedule for snapshots on the
tier2 disk providing you have enough of it.

Best practices in cases like this are completely based on your
business needs.

~F

Thanks... I take it then that it is generally a good practice to, in
some way or another, use the tier 2 as a stage prior to tape?
(assuming we have enough space on tier2)

Not necessarily. tier2 is great for seldom used data, or archive
data, or whatever else you can think of such that it doesn't need to
be on expensive tier1 disk.

~F

Right... I meant that in addition to staging...

I guess I should rephrase my original question... from your reply to
an earlier query of mine, which was, should I go from tier1 to VTL or
tape or should I go from tier1 to tier2 first (either via snap or
ndmp) prior to vtl/tape, I gather that you prefer the later...
correct?

Thanks...
Dvy

Backup data format
 

On 6 Feb 2007 17:08:26 -0800, wrote:


Not necessarily. tier2 is great for seldom used data, or archive
data, or whatever else you can think of such that it doesn't need to
be on expensive tier1 disk.

~F

Right... I meant that in addition to staging...

I guess I should rephrase my original question... from your reply to
an earlier query of mine, which was, should I go from tier1 to VTL or
tape or should I go from tier1 to tier2 first (either via snap or
ndmp) prior to vtl/tape, I gather that you prefer the later...
correct?

Thanks...
Dvy

It still depends on what you need. If the load placed on the tier1
disk is noticeable or impacts your users then offloading to tier2 is a
good idea.
If the load is acceptable then use the tier2 for other stuff.

There's no reason to complicate matters if there's no business need.

Personally I prefer to keep it simple, go direct from tier1. However,
business drivers are such that I have to offload the impact to tier2.
So I use Snapvault to tier2 then to tape.

~F