NoSQL Shake-Up. Membase and CouchOne merge into Couchbase
The companies describe the merger as a real synergy or optimal fit. Membase will enhance CouchDB performance by providing an efficient, scalable and distributed caching layer and speeding up view server operations. CouchDB replaces the Membase persistence store (SQLLite) and enhances Membase with querying, indexing, map-reduce and more database capabilities. It also benefits from the more mature operations and tools support that Membase offers. The combination of both allows the new products to serve many more different customer needs than before and scale up to millions of users or down to a single mobile device.
Q: The merger sounds like an well fitting synergy between both products. What about company culture and development approaches (as well as implementation languages)? How do you plan to merge those? Or will the products be developed separately and then merged at the product level?
Q: How would you describe the future product complexity - internal and perceived?
Membase customers were already storing documents (up to 20MB) in the key-value store. What they were asking for were better, structured means to query, index and work with the data contained in the documents. CouchDB customers on the other hand were asking for more scalability, performance, better operations and tooling support. So both sides benefit immensely. Regarding perceived complexity - it is only as complex as you need to know for your problems.
Q: How do you plan the transition to Couchbase, especially for developers and ops?
As Couchbase is targeted to be a Membase drop-in replacement there is nothing new for customers to learn, except if they want to use the advanced querying and synchronization features offered by CouchDB.
The CouchDB http protocol is the main protocol for Couchbase, so the same applies to CouchDB users. They only have to deal with Memcached configuration if they need to scale out.
Integration will happen step by step. First the products are integrated at the protocol level. That won't add much complexity because the layers are well isolated. Later, if communication performance reasons dictate it there will be a tighter, code level integration.
Q: As the clear focus of the previous products is merged together - what is the new tagline that people should associate Couchbase with?
The main tagline is "simple, fast, elastic" - even if this is a bit airy. But there is much more to it, especially the choice from a family of products that scale up and down.
Q: Regarding the different layers, one could say that CouchDB (mobile) is the offline cache of the online Membase cache for CouchDB?
Actually the Membase product manager Frank Weigel likes to call it like that, but has got to argue with CouchDB's Jan Lehnardt who doesn't support that notion.
Q: What is the state of mobile CouchDB ?
Erlang for running CouchDB natively on the mobile platforms was easy to port to the Android kernel but it took much more effort to achieve that in iOS. But the iOS version will be released soon, then mobile applications will follow that use this embedded CouchDB as a synchronizable, offline document store.
Q: Is the merge between a document database and a key value store creating a new product category?
We would call it high performance document database, but it can also be used as high performance caching layer, so it might be indeed a new product category. But mostly it is about solving problems for customers.
Q: What is the Timeline for the integrated product?
We have an aggressive timeline. Within 30 days the product families will be presented together on a single web site - couchbase.com. With several public alpha and beta stages we strive to have a first release of an integrated system with CouchDB as the backend for memcached by summer. By the end of the year there will be a complete integrated product family that works internally with a shared protocol.
Q: Where will the company be located, will the teams be co-located?
Our teams are already distributed world wide. It is too hard to hire excellent people in one place. So we continue the distributed collaboration that depends heavily on the ability of our people to be excellent communicators and also by using easy lines of communication like Skype, IM, IRC.
Q: Do you think similar market aggregation will happen in other areas of the NoSQL space? What is your prediction for its development in 2011?
We don't like to look into the crystal ball. All NoSQL players talk to their customers and try to solve their problems. We all get similar feedback about the need for solutions that can solve more than one use-case. So there is a needed convergence as well as the power of economics that come into play here. There will be more consolidation and complete product lines.
Q: What about the "Choice" part/polyglot persistence in NoSQL do you put that to an end, is Couchbase a "one size fits all" solution, the next generation Oracle? What is your stance on that?
No, we are not afraid of becoming the next Oracle, actually that would be a luxury problem to solve. There will be always niches for certain special purpose approaches, we think there will be a few general purpose NoSQL databases. In the future it is possible that Couchbase joins forces with other NoSQL providers. Polyglot persistence will continue but the focus will shift to cost effectiveness and extreme scalability, and mobile will get more and more important.
Thanks so much for taking the time answering our questions.