Discussion:
Question regarding RDC and large file
(too old to reply)
Bob Williamson
2006-04-28 16:09:54 UTC
Permalink
We run Sharepoint 2001 (soon to be SharePoint 2003). I have a batch file
that backs up the SharePoint store and then zips it and overwrites the
backup from the night before. The SharePoint document library consists of
about 90,000 documents and is about 2gb zipped.

My questions is, will the RDC engine be able to only send the differences
with a file like this?

Thanks,
Bob
Bob Williamson
2006-04-28 16:19:33 UTC
Permalink
One more thing. Currently the zip file gets replaced, but I could modify it
to add the backup file to the zip (as opposed to replacing the entire zip).
The zip contains one large file "Backup.bak"

Thanks,
Bob
Post by Bob Williamson
We run Sharepoint 2001 (soon to be SharePoint 2003). I have a batch file
that backs up the SharePoint store and then zips it and overwrites the
backup from the night before. The SharePoint document library consists of
about 90,000 documents and is about 2gb zipped.
My questions is, will the RDC engine be able to only send the differences
with a file like this?
Thanks,
Bob
Dave Mills
2006-04-29 12:52:56 UTC
Permalink
I would think that almost every element of the zip file gets changed. If you add
a 20k file at the start then the zip file will start with that file and this
will displace the data for ever other part of the zip file. Appending only
modified data may help leave the bulk intact so that DRS can take advantage of
that.

data would On Fri, 28 Apr 2006 09:09:54 -0700, "Bob Williamson"
Post by Bob Williamson
We run Sharepoint 2001 (soon to be SharePoint 2003). I have a batch file
that backs up the SharePoint store and then zips it and overwrites the
backup from the night before. The SharePoint document library consists of
about 90,000 documents and is about 2gb zipped.
My questions is, will the RDC engine be able to only send the differences
with a file like this?
Thanks,
Bob
--
Dave Mills
There are 10 type of people, those that understand binary and those that don't.
Bob Williamson
2006-05-02 00:12:53 UTC
Permalink
Yeah, that's what I was thinking as well.

In theory, if I backup the portal server on top of the prior desktop the
engine should only transfer the delta. Since Sharepoint runs in an exchange
type DB, does that affect more of the backup file? A Backup of a sharepoint
portal is peformed via a VBS msdmback.vbs. I THINK it exports the data to a
flat file. Thus the delta might be small.

If I choose to "add" the file to the existing zip, the process should ask if
I want to overwrite the existing file in the zip. In affect overwriting the
existing backup. BUT will the zip delta be huge because the process of
zipping.....that is the question.

Then again I coudl simply not zip the 10gb backup file and try using the RDC
engine to just transfer that delta.....

I guess it all comes down how Sharepoitn 2001 backups and whether the engine
can distignuish the differences between the old and new backups.

Bob
Post by Dave Mills
I would think that almost every element of the zip file gets changed. If you add
a 20k file at the start then the zip file will start with that file and this
will displace the data for ever other part of the zip file. Appending only
modified data may help leave the bulk intact so that DRS can take advantage of
that.
data would On Fri, 28 Apr 2006 09:09:54 -0700, "Bob Williamson"
Post by Bob Williamson
We run Sharepoint 2001 (soon to be SharePoint 2003). I have a batch file
that backs up the SharePoint store and then zips it and overwrites the
backup from the night before. The SharePoint document library consists of
about 90,000 documents and is about 2gb zipped.
My questions is, will the RDC engine be able to only send the differences
with a file like this?
Thanks,
Bob
--
Dave Mills
There are 10 type of people, those that understand binary and those that don't.
Robert A Post, Jr (MSFT)
2006-05-02 20:47:18 UTC
Permalink
This depends a lot on the application zipping the file. If it uses a block
level compression then you may see a lot of RDC savings, otherwise most of
the zipped file will be changed when new contents are added. I would
suggest trying the no zip approach, if you have space. DFSR will compress
the data for file transfer using it's own block level compression and if it
is a incremental backup then most of the data should be common and RDC
should do very well.

Thanks,

Robert A Post, Jr [MSFT]
Distributed File System Replication Test
This posting is provided "AS IS" with no warranties, and confers no rights.
Bob Williamson
2006-05-03 00:47:27 UTC
Permalink
How can I tell if the RDC is working well? Can I turn on a log or
something?

Thanks,
Bob
Post by Robert A Post, Jr (MSFT)
This depends a lot on the application zipping the file. If it uses a block
level compression then you may see a lot of RDC savings, otherwise most of
the zipped file will be changed when new contents are added. I would
suggest trying the no zip approach, if you have space. DFSR will compress
the data for file transfer using it's own block level compression and if it
is a incremental backup then most of the data should be common and RDC
should do very well.
Thanks,
Robert A Post, Jr [MSFT]
Distributed File System Replication Test
This posting is provided "AS IS" with no warranties, and confers no rights.
Robert A Post, Jr (MSFT)
2006-05-03 02:06:28 UTC
Permalink
You could use any of these methods, depending on the granularity of data
you want. The first would be the best way to determine the savings for a
single file.

1. Scour the DFSR debug logs (%systemdrive%\windows\debug\DFSR*.log)
2. Check the perf counters for the "RDC*" perf counters for a specific
replicated folder through perfmon
3. Run a health report through the DFS mgmt snapin

Thanks,

Robert A Post, Jr [MSFT]
Distributed File System Replication Test
This posting is provided "AS IS" with no warranties, and confers no rights.
Continue reading on narkive:
Loading...