Strategy for offsite IFW backup data replication?

[robita]

I am running IFW complete system backup on a Windows 2K3 server twice a day via system scheduler.
The backups land on an SMB share of a dedicated linux-centos backup server with 2Tb storage.
One IFW backup is to a full size data file, approx 80Gb and the other splits into 4Gb chunks.
Seven days of complete system backups are held on the linux system which auto housekeeps via cron.
This setup has been working fine for over a year and the couple of times we had to restore, there were no problems.

Now, insurance requirements dictate we must have a copy of the complete system backup held offsite, in case of disaster.
I have an identical linux-centos backup server at an offsite location and would like to replicate the onsite backup data to it.
I am thinking about using sshfs to sync the data between the two backup servers via DSL.

Obviously, attempting to copy 80+Gb of data via 8Mb/s DSL every day is not practical.
Therefore, I need to re-think the whole strategy to make this work properly.

I realise that some implementation of differential/incremental backup is likely the way to go.
So far, I only have experience of using IFW for a complete system backup.

Bearing in mind there are currently two new complete system backup sets every day;
I am struggling to grasp the concept of how / what to implement to ensure the offsite backup server always contains an (almost) up to date complete system backup which is available to restore if/when necessary.
For example, in case of disaster, we may need to restore the offsite backup to new hardware for replacement on site.

Could somebody please provide some guidance on how this solution may be achieved?
Many thanks in advance.

[TeraByte Support]

I know some do it via uncompressed options and differentials and use rsync
to keep files up to date. The advantage to uncompressed is the data doesn't
change as much and able to sync a bit better.

"robita" wrote in message news:2570@public.image...

I am running IFW complete system backup on a Windows 2K3 server twice a day
via system scheduler.
The backups land on an SMB share of a dedicated linux-centos backup server
with 2Tb storage.
One IFW backup is to a full size data file, approx 80Gb and the other splits
into 4Gb chunks.
Seven days of complete system backups are held on the linux system which
auto housekeeps via cron.
This setup has been working fine for over a year and the couple of times we
had to restore, there were no problems.

Now, insurance requirements dictate we must have a copy of the complete
system backup held offsite, in case of disaster.
I have an identical linux-centos backup server at an offsite location and
would like to replicate the onsite backup data to it.
I am thinking about using sshfs to sync the data between the two backup
servers via DSL.

Obviously, attempting to copy 80+Gb of data via 8Mb/s DSL every day is not
practical.
Therefore, I need to re-think the whole strategy to make this work properly.

I realise that some implementation of differential/incremental backup is
likely the way to go.
So far, I only have experience of using IFW for a complete system backup.

Bearing in mind there are currently two new complete system backup sets
every day;
I am struggling to grasp the concept of how / what to implement to ensure
the offsite backup server always contains an (almost) up to date complete
system backup which is available to restore if/when necessary.
For example, in case of disaster, we may need to restore the offsite backup
to new hardware for replacement on site.

Could somebody please provide some guidance on how this solution may be
achieved?
Many thanks in advance.

[robita]

Ok, so uncompressed and differentials is the way to go.
I should be ok with the rsync part but need some help with the rest of it.
Can you provide any examples or links to how-tos?
Many thanks.

[robita]

Following some housekeeping, I managed to reduce the full system backup to 65Gb.
Now I am doing an uncompressed differential each day, it is 30Gb.
This is still not practical to rsync over DSL every day.
Any ideas, please ?

[TeraByte Support]

why is there 30g of changes? You should be using the latest version and not
be defragging everyday. I believe they were doing uncompressed so rsync can
sync the data that already existed more easily and compress on the transfer.

"robita" wrote in message news:2737@public.image...

Following some housekeeping, I managed to reduce the full system backup to
65Gb.
Now I am doing an uncompressed differential each day, it is 30Gb.
This is still not practical to rsync over DSL every day.
Any ideas, please ?