Looking for suggestions on repository design

I am looking for some ideas here. We have an ERP product at work that we get
the source for. We also make some modifications to the source to add some of
our own business processes in to the application. So my first question would
be how would you recommend laying out the stucture for something like that?
I was thinking about the trunk being the unmodified source, then the modifed
source being a branch. What are your thoughts?

Also, the next big problem/question. All of the source is delivered directly
to our server via patches and upgrades. So then I would have to download the
new version from the server to my client, then upload to svn. I don't care
about tracking changes in the delivered source as much, but I would like to
make it easy to merge the latest delivered source to our customizations so
we don't have to manually patch or reapply all our modifications as we have
quite a few programs with many modifications. Any suggestions on how to make
this work? Can I merge two files with the same name, but have different
histories?

Note that I haven't tried any of this yet and I am just starting to learn
svn. But mostly I am seeking best practices from those who have had these
expiriences. Any feedback would be appreciated.

-- 
Mike Wills
Midrange Programmer/Lawson Software Administrator
koldark@gmail.com
http://mikewills.name
Want Gmail? Email koldark+gmail@gmail.com to get on my waiting list.