[<<]Message[>>]    [<<]Author        [<<]Subject[>>]    [<<]Thread[>>]

Number   :  4207
Date     :  2003-04-05
Author   :  agressiv99
Subject  :  Re: Delta-Byte Replication
Size(KB) :  2


I see what you are saying - So I would take these large files which 
change just a bit each day, copy them to a directory. Copy those 
files to the remote location, and then the next day, run an xdelta 
locally from the directory the users work from, and the local archive 
directory that I had just created. Then, copy the "diff" file up to 
the remote server and apply the diff locally each day.

So you are right, do I have space for 2 copies of each of these 
files? That is the question I'll have to dive into. Volume Shadow 
Copy would eliminate this need in Windows Server 2003, but I am not 
aware of a command-line method of getting to previous versions of the 
volume shadow; you can only use the GUI, which wouldn't help here.

I'd have to run scripts remotely on each of these machines, but you 
are right, it makes it feasable. Now I just have to weigh that 
versus spending about $10,000 on our sites where using a plain 
copy/robocopy/xxcopy on its own is not feasible. 

Thanks -

Greg


--- In xxcopy@yahoogroups.com, "J. Merrill"  wrote:
> Can you afford the space on your local system to keep a copy of the 
file as it is at the other end of the WAN link? (The "local copy" 
could also be on your local network, as when you did your testing.)
> 
> If you can, you could build the xdelta file based on the local 
copy; apply the xdelta file to the local copy; compare the current 
real file with the local copy and if they don't match exactly 
(meaning that xdelta didn't do exactly what it should have done) then 
re-copy the current real file to your local copy and also over the 
WAN. If the xdelta process worked perfectly, send the xdelta file 
over the WAN and apply it on the other end.
> 
> One advantage of this approach is that a bug in the xdelta program 
would not cause you to have an incorrect copy of the file on the 
other end of the WAN link (assuming that the "apply xdelta" program 
behaves identically on both systems).
> 
> Good luck...
> 
> At 04:05 PM 4/2/2003 +0000, agressiv99 wrote
> >Thanks for the response -
> >
> >I gave a win32 port of xdelta a try. Locally, it works great. 
> >However, going over the wire, it is another story.
> >
> >Example. I have a 134MB file on one computer, connected at 10MB 
> >ethernet to another.
> >
> >Copying the file takes about 90 seconds.
> >
> >changing about 10 bytes in the file and running an xdelta takes 
about 
> >140 seconds and uses a ton of bandwidth which is what I was trying 
to 
> >avoid. I can't imagine it over a 256k frame relay.
> >
> >I guess without the program running on both ends and talking to 
each 
> >other like Storage Replicator or NSI Doubletake, I don't see a 
> >command-line solution to it. Unfortunately those (as well as any 
> >internet-based solutions) require a ton of cash for an environment 
> >our size.
> >
> >Greg
> 
> 
> J. Merrill / Analytical Software Corp



This message if part of XXCOPY's message Archive.
The archive contains all the messages posted at Yahoo!Groups: XXCOPY.