Jim stated:

> But let me repeat my question. Why do you need to preserve the
> records at the "freeze point"? Assuming the rest of the problems
> have solutions, the records that exist at the freeze points are the
> only records you don't need to produce an incremental backup. What
> is the gain in trying to preserve them?

<for sake of discussion, this is the solution looking for problems again...
my apologies>

Depending on what problem you are solving, you may or may not need them. The
problem that I was stating would not need them as long as there was a backup
as of that point in time and you didn't have the requirement of ever needing
to produce that same backup image again or rolling back the database to
reflect it's prior state.

However, if in my problem I had required a differential, rather than an
incremental, backup it may become necessary to allow more than one freeze
point at a time in the database. If this were to become a requirement then a
way to tag a BDR as a named transaction member would need to be established.

What would probably alleviate the need of this and allow those records to be
dropped is if it were made possible to incrementally move the new changes
into an external log file (which essentially is a journal of sorts I
suppose). Once stuff is appended into the external incremental backup file
the internal OANT (oldest active named transaction) could be up'd to the
point being ran from. In this way you could have a total backup once a week
and then maintain a delta file that would be one contiguous differential
backup file incrementally appended to on say a daily or perhaps even and
hourly basis.

This provides a uniform way to handle pieces of a database's work activity
in a super transaction that allows a DBA to do a lot of creative things to
solve backup and restore problems and crude replication problems. This
certainly would be sufficient to cover the problem that I have at hand.

FWIW,
Jason Wharton
CPS - Mesa AZ
http://www.ibobjects.com