[CODE4LIB] CouchDB and MongoDB (was: Re: O'Reilly books...)
On Tue, Dec 14, 2010 at 11:58 AM, Bill Dueber b...@dueber.com wrote: Oops. I just found a better overview than I can provide, at http://www.mongodb.org/display/DOCS/Comparing+Mongo+DB+and+Couch+DB I was just about to send that link. There are lots of other players in this space, too -- see http://nosql-database.org/ It depends on how you define that space. There are lots of players in the non-relational, AKA, NoSQL space, but in the document oriented space I don't know of any other current contender other than MongoDB and CouchDB. Do you? Comparing Riak, Cassandra and MongoDB is like comparing a golf cart, a fork lift and a fire engine. They are just too different. But i'd say MongoDB and CouchDB belong in the same category, though MongoDB is optimized for performance in cluster, deployed in a single datacenter, with master-slave replication, and CouchDB is designed for easy and reliable distributed deployment with master-master replication among nodes that are not always online. Their conceptual data model is very similar (JSON and BSON), so it's a snap to migrate data from CouchDB to MongoDB (the opposite maybe more complicated depending on the dataset, because BSON has more primitive types than JSON). Where I work [1] we are doing pilot projects with CouchDB, but we also envision using CouchDB as the main repository for content creation, and pushing data to MongoDB for high demand services, if we find out that CouchDB can't handle the traffic. [1] http://regional.bvsalud.org/php/index.php?lang=en -- Luciano Ramalho programador repentista || stand-up programmer Twitter: @luciano
Re: [CODE4LIB] CouchDB and MongoDB (was: Re: O'Reilly books...)
Tongue lodged deeply -- so deeply -- in cheek: http://nosql.mypopescu.com/post/1016320617/mongodb-is-web-scale# NSFW if your co-workers don't like to hear computer-generated swears. :) Cheers, -Nate On Tue, Dec 14, 2010 at 11:11 AM, Luciano Ramalho luci...@ramalho.org wrote: On Tue, Dec 14, 2010 at 11:58 AM, Bill Dueber b...@dueber.com wrote: Oops. I just found a better overview than I can provide, at http://www.mongodb.org/display/DOCS/Comparing+Mongo+DB+and+Couch+DB I was just about to send that link. There are lots of other players in this space, too -- see http://nosql-database.org/ It depends on how you define that space. There are lots of players in the non-relational, AKA, NoSQL space, but in the document oriented space I don't know of any other current contender other than MongoDB and CouchDB. Do you? Comparing Riak, Cassandra and MongoDB is like comparing a golf cart, a fork lift and a fire engine. They are just too different. But i'd say MongoDB and CouchDB belong in the same category, though MongoDB is optimized for performance in cluster, deployed in a single datacenter, with master-slave replication, and CouchDB is designed for easy and reliable distributed deployment with master-master replication among nodes that are not always online. Their conceptual data model is very similar (JSON and BSON), so it's a snap to migrate data from CouchDB to MongoDB (the opposite maybe more complicated depending on the dataset, because BSON has more primitive types than JSON). Where I work [1] we are doing pilot projects with CouchDB, but we also envision using CouchDB as the main repository for content creation, and pushing data to MongoDB for high demand services, if we find out that CouchDB can't handle the traffic. [1] http://regional.bvsalud.org/php/index.php?lang=en -- Luciano Ramalho programador repentista || stand-up programmer Twitter: @luciano
Re: [CODE4LIB] CouchDB and MongoDB (was: Re: O'Reilly books...)
On Tue, Dec 14, 2010 at 3:23 PM, Nate Vack njv...@wisc.edu wrote: Tongue lodged deeply -- so deeply -- in cheek: http://nosql.mypopescu.com/post/1016320617/mongodb-is-web-scale# Yeah, that is funny, thanks for the link, Nate. I bet you did not mean any harm, but I hope the joke does not kill the conversation we had just started on the other thread. -- Luciano Ramalho programador repentista || stand-up programmer Twitter: @luciano
Re: [CODE4LIB] CouchDB and MongoDB (was: Re: O'Reilly books...)
On Tue, Dec 14, 2010 at 5:57 PM, Tom Keays tomke...@gmail.com wrote: I saw this visualization of where the various nosql databases fit on the CAP Theorem triangle. CAP says there are three primary concerns you must balance when choosing a data management system: Consistency, Availability, and Partition tolerance. Furthermore, you can only pick 2. http://blog.nahurst.com/visual-guide-to-nosql-systems According to this, Riak, SimpleDB, Cassandra and CouchDB all sit on the AP side, whereas MongoDB and BigTable sit on the CP side. Most relational databases sit on the CA side. Very interesting, Tom, thanks for the link. It is interesting to note that although CouchDB and MongoDB sit on different sides of the CAP triangle, their data model, from an application perspective, is very similar. But the implementation of the data model is very different, as Bill Dueber mentioned before, with MongoDB doing updates in-place whenever possible, and aggressively caching writes, both of which increase update speed but also the risk of a corrupt database in case of a crash. CouchDB does neither, so updates are much slower, but its data is always in a consistent state on disk, because it only appends, and appends are guaranteed to be atomic in posix systems. -- Luciano Ramalho programador repentista || stand-up programmer Twitter: @luciano
[CODE4LIB] CouchDB
Hey all, I just started experimenting with CouchDB the other day, and it's pretty cool. With the amount of data the Botanical Heritage Library (BHL) is carrying, this may be an option for the future. Does anyone have any experience with it, or any pointers to a good howto, or basic setup/usage case? I appreciate that it's a different approach to an age old problem, and I can see it working hand in hand with things like hadoop (hdfs), lucene/solr, etc. P
Re: [CODE4LIB] CouchDB
Hi Phil, I'm starting to play with CouchDB myself, mainly as a way of learning about schemaless databases. Have you seen the book that is being written, http://books.couchdb.org/relax/ ? So far it's got a pretty good set of up-and-running instructions and some basic howtos. Mark - phil cryer p...@cryer.us wrote: Hey all, I just started experimenting with CouchDB the other day, and it's pretty cool. With the amount of data the Botanical Heritage Library (BHL) is carrying, this may be an option for the future. Does anyone have any experience with it, or any pointers to a good howto, or basic setup/usage case? I appreciate that it's a different approach to an age old problem, and I can see it working hand in hand with things like hadoop (hdfs), lucene/solr, etc. P
Re: [CODE4LIB] CouchDB
On Tue, Mar 17, 2009 at 7:22 AM, phil cryer p...@cryer.us wrote: Hey all, I just started experimenting with CouchDB the other day, and it's pretty cool. With the amount of data the Botanical Heritage Library (BHL) is carrying, this may be an option for the future. Does anyone have any experience with it, or any pointers to a good howto, or basic setup/usage case? I appreciate that it's a different approach to an age old problem, and I can see it working hand in hand with things like hadoop (hdfs), lucene/solr, etc. For full-text search, some experimental work has been done. There's hypercouch which brings Hyper Estraier and CouchDB together: http://github.com/davisp/hypercouch/tree/master There's also couchdb-lucene which uses Lucene for full text search: http://github.com/rnewson/couchdb-lucene/tree/master People are still exploring this uncharted land (so to speak). Querying is accomplished by hooking up an external service to CouchDB. http://wiki.apache.org/couchdb/ExternalProcesses This is basically a process that stays resident, reads requests on STDIN and sends responses on STDOUT as the wiki page I linked to above describes. As for indexing, I think the smart way to do it is to follow couchdb-lucene's example and setup an update_notification script. --beppu