incubator-couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michael Ludwig <mil...@gmx.de>
Subject Re: About denormalization and keep consistent
Date Fri, 09 Apr 2010 20:30:37 GMT
Andrew Melo schrieb am 09.04.2010 um 15:04:59 (-0500)
[Re: About denormalization and keep consistent]:

> > Are there scenarios where writing happens to an SQL server, from
> > where NoSQL replicas are fed and used to serve readers?
> 
> Possibly. But, if you're writing to a single SQL master, then you lose
> two benefits of Couch: horizontally scalable writes and not having to
> be restricted by SQL's data model. You could do (for instance) single
> master writes and many-slave reads on a stock SQL server in that case.

Thanks, Andrew. It appears Faust is concerned about data consistency
which is usually and easily ensured by suitable SQL data modeling. That
does not happen to be the forte of a system designed for scalability at
the expense of consistency.

Faust, there are many places and cases where data can be copied and held
redundantly in hundreds or thousands of places, and it simply doesn't
matter. Think email, for example. If you changed the name belonging to
your email address to "Johann Faust", it wouldn't matter that there are
mails with the older version of your name in the archive, and they
wouldn't have to be updated, and they even shouldn't be updated. That
seems to be an excellent use case for a system like CouchDB.

If, on the other hand, you care about consistency in the way you
described, you might want to consider using a system that supports
you in guaranteeing such consistency.

-- 
Michael Ludwig

Mime
View raw message