Re: Future In-Core Replication

Robert Haas <robertmhaas@gmail.com> wrote:
> The other half of the changes - applying the updates - is
> relatively straightforward, and it wouldn't bother me to leave
> that in user-land, especially in the MMR case, where you have to
> deal with conflict resolution rules that may be much simpler to
> express in a higher-level language than they would be in C.
I've developed successful MMR more than once, and while I wouldn't
say it's exactly been implemented in the database, it hasn't exactly
been in application space either.  The most successful
implementations I've worked with have been a layer just outside the
database, of which application software was completely unaware.  The
database wasn't aware of the coordination per se; but it *did* need
to provide transaction information in a clean way, and the
declarations of how data was distributed were in the database.  In
my experience a declarative definition of data distribution has
always been sufficient, and certainly cleaner to deal with than
imperative coding would be.
YMMV.
-Kevin