couchdb-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Chris Anderson <jch...@apache.org>
Subject Re: Replication and forms of weak consistency
Date Mon, 16 Feb 2009 05:25:23 GMT
On Sun, Feb 15, 2009 at 8:48 PM, Antony Blakey <antony.blakey@gmail.com> wrote:
> It would suffice to say "don't presume that just because you update A before
> you update B your code will see A if it can see B". This is the essence of
> not having Monotonic Writes. It's not about transactions at all. Isolated
> write groups are something that would a) optimise replication; and b) be
> useful to application developers (especially with a fail-on-commit option),
> but they aren't transactions in the traditional sense.

So it seems as though, when a long history is replicated under your
model (interleaving many different client updates) we would end up
sending a lot more data over the wire under your proposed model. In
order to ensure that the isolation group stays together, even should
replication fail before completion, we'd have to send the latest
doc-rev for every doc touched in each isolated doc group.

In the current system we just send the latest non-conflicted rev or
all the conflict revs is they exist. It makes for a lot less data on
the wire. (Correct me if I'm wrong.)

Your story about comments being replicated without their assocaited
posts is a good example of the counter-intuitive things that can
happen when replication fails before completion. Thanks for that.

I think these questions are interesting, I really do. However, in my
mind, what makes CouchDB relaxing, is that we're not trying to be
ambitious on the transactional guarantees front. So far, we've tried
to give only the guarantees we know we can afford to give, and
concentrate on getting them right.

I certainly don't want to squelch any innovation here (and please keep
discussing the topic). I just think that CouchDB's bare-bones
guarantees are more powerful than we're giving them credit for.
Robert's point that much of this can be implemented on top of CouchDB
is an interesting one. If it is indeed the case, then the question
becomes whether clients or the database should be responsible for
providing the transactional API.

-- 
Chris Anderson
http://jchris.mfdz.com

Mime
View raw message