It's an interesting release.. I had been tinkering with the beta. Oddly enough (considering the name) to me it seems really more tempting as a replacement for memcached than for couchdb.
Can someone from Couchbase or someone familiar with their plans please confirm the points here:
* Basic data storage operations must use the memcached API.
* However, the HTTP REST API for basic CRUD operations must be updated to use the memcached protocol.
It seems that the couchdb REST API is no longer supported. Was that due to resource constraints in getting 2.0 shipped or is it an intentional design decision to not support this?
It's not so much that you need to use the memcached API. We've built a set of client libraries that give you a consistent API (check out couchbase.com/develop) though parts are done through memcached binary protocol (for speed) and parts are done through a RESTful interface inspired by CouchDB.
It wasn't about resource constraints; it that we felt going after distributed deployments and having the speed to keep up with application demands (mostly driven by CRUD) was important. That's what devs and ops people kept telling us they'd need.
Can someone explain to me in a few words what the difference is now between CouchDB, Membase, Couchbase?
I was looking into CouchDB a while back and loved its concepts but then it became an apache project and there was couchone and membase and coucbase and i lost track of that is happening in the community.
In short: Is Couchbase somewhat like the next CouchDB? Does it still use HTTP/JS everywhere? I loved the fact that i could stay in JS/HTTP/HTML in CouchDB, query CouchDB directly from the browser and even build a whole website in that system. Now it looks like there are a lot of client libraries again, Is this not HTTP anymore?
CouchDB: Still there, still doing their thing but now as an Apache project. A few companies have built businesses around it such as IrisCouch (SAAS platform) and Cloudant (who maintain the open-source BigCouch which is API-compatible with CouchDB but adds more scalable clustering).
CouchBase 2.0 (née Membase/CouchOne): Basically a version of MemBase (which was developed by the memcached developers) with some of the engine parts and concepts from CouchDB mixed in. My understanding is that a lot of things were also written from scratch. The idea is that it's the best of both worlds: good in memory dataset performance, disk persistence, data replication, cluster rebalancing and partitioning. Not sure how it works in practice.
The name is a bit unfortunate because it appears to be more like memcached from the developer API perspective than CouchDB. If successful, I don't think CouchBase will replace CouchDB but it will instead carve out its own place in the ecosystem.
Apache CouchDB is the original opensouce project. The guy who started it left a year ago (?), but the project is alive. How much so? I don't know..
Membase does not exists, it was renamed to Couchbase Server. It used to be CoucbDB + Memcache, which seems like a strange combo to me, but I haven't tried it. Also, I guess it has evolved since then. According to wikipedia: "It has a separate and independent community, provides a very different set of capabilities, and supports very different use cases".
Thanks for the responses. In the end it's what i feared, there is some open source fork with relatively uncertain future and the new database by the original creator is a totally different beast.
Not saying CouchDB is dead but at the point in time i wouldn't feel comfortable recommending it :(
I wouldnt worry, Damien hadnt been particularly active with CouchDB dev for quite a while before CouchBase happened.
CouchDB has a new release pending, a bunch of new features and work in development.
The CouchBase situtation certainly caused a lul in CouchDB activity, mostly due to the confusion over the branding, but it has most definitely gotten through the other side of that.
What's your worry about an open source fork? Apache CouchDB is in solid shape and we at Cloudant are working to merge the BigCouch clustering layer back into Apache CouchDB for the 2.0 release.
We've been looking at Couchbase for our startup, and have found it a great fit for our needs. Unfortunately, we needed the functionality of 2.0, which wasn't available at the time. Since then we've been building on Postgre. Seems we'll have to take it into consideration again...
Do you know if a client for the Go language will be released?
edit: Sorry, I see there's a Go client as a community SDK.
Looks interesting, too bad they don't provide an official node.js client, as the unofficial one doesn't seem to be widely used so it's a bit risky (I guess there's a reason it's called 'experimental')
I worked on the node client. The core API is mostly stable, so it should be fine to work with. We are still adding some of the less-used functionality, but for basic operations, I've been using it and its fine. Github link https://github.com/couchbase/couchnode
It's a pity their ruby client does not support jruby since it directly depends on a c lib. I guess you can go and use the java client and wrap it, but it would be nice to have a drop-in replacement.
Can someone from Couchbase or someone familiar with their plans please confirm the points here:
http://www.couchbase.com/docs/couchbase-manual-2.0/couchbase...
Specifically these points:
* Basic data storage operations must use the memcached API.
* However, the HTTP REST API for basic CRUD operations must be updated to use the memcached protocol.
It seems that the couchdb REST API is no longer supported. Was that due to resource constraints in getting 2.0 shipped or is it an intentional design decision to not support this?