17

I keep seeing references to the idea that "CouchDB may not be the best tool in every situation." This is good to know, but unfortunately also applies to every technology.

What would be much more helpful is a description of how CouchDB was tried on a project and subsequently abandoned for a traditional SQL database.

If you've tried CouchDB on a project and later gone back to a SQL database, what factors played the biggest role?

Paul D. Waite
  • 96,640
  • 56
  • 199
  • 270
Rich Apodaca
  • 28,316
  • 16
  • 103
  • 129

5 Answers5

13

You might want to check out the following articles, which describe a few examples:

Community
  • 1
  • 1
Daniel Vassallo
  • 337,827
  • 72
  • 505
  • 443
  • 1
    @Daniel, those are some useful links, thanks. But I didn't find any discussions of a specific project where the limitations/mismatch in CouchDB were so great that it forced the project to go back to a RDBMS. That's really what I'm looking for. – Rich Apodaca Jan 12 '10 at 20:41
  • 1
    @Rich: I wouldn't look for examples that chose CouchDB and were forced to go back, because in some situations, a relational database, with traditional transactions and queries is the only option. Recently I came across an Oracle case study on Betfair.com http://www.oracle.com/customers/snapshots/betfair-database-case-study.pdf. I think that with 20m transactions per day, dealing with peaks of 1k transactions per second with real money, operating under strict regulations, while managing to serve over 500 million page view per day, they are doing an excellent leverage on relational databases. – Daniel Vassallo Jan 12 '10 at 21:27
4

There’s a good CouchDB post-mortem from Sauce Labs here:

And another from Signal here:

Paul D. Waite
  • 96,640
  • 56
  • 199
  • 270
4

Its marketing slogans "Relax, it's easy" and "CouchDB bult for the Web" - it's not true.

Because:

  1. It's not easy. Ha-ha, writing map/reduce for every query is easy, really? Did someone who advocates it actually tried this approach, on something real I mean, not just two-weekends-blog app?
  2. Some common web-app tasks is hard to do with CouchDB (try to get Post with its Comments Count in one query or Tag Cloud (top N tags, not just counts of tags) to see it for Yourself.

So, CouchDB has very interesting and unique features, but its marketing is wrong. It's not for a general web app, it's a nice but a very specialized tool.

P.S. More details http://alex-craft.com/blog/2013/a-little-about-cochudb-and-comparison-with-mongodb

Alex Craft
  • 13,598
  • 11
  • 69
  • 133
  • The question was "Are there any documented projects where CouchDB was tried and rejected?" You haven't mentioned any such projects. – Paul D. Waite Jul 26 '13 at 09:26
  • You are right, in my message I implied that I tried to use it in my project and rejected it (switched to MongoDB) because common things needed by common web app was too hard to do. – Alex Craft Aug 16 '13 at 20:47
  • ah, yes I see. Looks like your blogpost is offline at the moment (I can’t access it anyway). – Paul D. Waite Aug 17 '13 at 17:57
3

you might find this page useful http://www.mongodb.org/display/DOCS/Comparing+Mongo+DB+and+Couch+DB

dm.
  • 1,982
  • 12
  • 7
2

Ubuntu stopped using CouchDB for UbuntuOne, because it failed to deliver its promises, even for small load: https://lists.ubuntu.com/archives/ubuntu-desktop/2011-November/003474.html

deronnax
  • 23
  • 6