[jimbokun]

One interesting thing about the legacy of CouchDB is the sync protocol. There are now multiple implementations capable of syncing with each other, making it easy to migrate data to a different provider, even if the underlying implementation is different. I think there have been one or two forks and different service providers, and they can all exchange data without issue, as far as I know.

[jchrisa]

HTML5 - http://pouchdb.com iOS, Android, .NET, Java - http://developer.couchbase.com/mobile Hosted - http://www.iriscouch.com and http://cloudant.com Original - http://couchdb.apache.org

All of these are 100% open source and interoperable.

[marpalmin]

This is exactly why I went for couchbase and decided to avoid firebase.

[skrebbel]

Assuming you use PouchDB, how do you avoid syncing the entire database with all users?

[willholley]

database-per-user is a common pattern with CouchDB. If you need to aggregate data across your users for reporting etc then you can use a server-side replication to create a "master" database which is only accessible to your internal users.

You could also use filtered replication but it is: a) not scalable b) not secure (the whole db would still be accessible to all users)

Couchbase has a different spin on this via "channels" though I'm not sure how it interoperates with PouchDB.

[jchrisa]

Couchbase Sync Gateway's channels are just slices through the data. The mechanism is internal to the Sync Gateway (you write a JS function that says who can see which channels and which channels a doc belongs to.) Replicating clients don't Need to know about this filtering, so channels work great with all flavors of Couch.