I'm not familiar with neo4j, but it appears they have full support for transactions: http://wiki.neo4j.org/content/Transactions You'd have to talk to the neo4j people to learn more about this.
Whether or not a document-oriented database and CouchDB specifically is a good fit for your model depends a lot on the operations and queries you'd like to perform, I think. CouchDB uses eventual consistency, so there is indeed no guarantee your data is consistent at any given moment. (The CAP theorem gives you availability and partition tolerance in return.) Nils. ________________________________________ Van: Ido Ran [ido....@gmail.com] Verzonden: zondag 19 december 2010 18:05 Aan: user@couchdb.apache.org Onderwerp: Re: Document model of network Hi, I have look at neo4j in the past, I now see they have RESTful interface. But, even there there is no mention of transaction on the RESTful interface, only CRUD operation on single entities which is something I already have. Do you agree in general that model with reference like link in graph which is not specifically belong to either side of the link is not a good candidate for document-oriented database? Ido On Sun, Dec 19, 2010 at 6:21 PM, Nils Breunese <n.breun...@vpro.nl> wrote: > I'm not sure if a document-oriented database like CouchDB is such a good > fit for your data model. Have you looked at graph databases like neo4j [0]? > > Nils. > > [0] http://neo4j.org/ > ________________________________________ > Van: Ido Ran [ido....@gmail.com] > Verzonden: zondag 19 december 2010 16:17 > Aan: CouchDB MailingList > Onderwerp: Document model of network > > Hi, > I have application which the primary data model is a network (graph). > Currently I use RDBMS in which I have table of Node and table of Link. The > Link table has FromNode, ToNode and LinkType fields. > > What is a good document model to put this type of data in CouchDB? > Should I provide information about my application that I didn't provide > here? > > Since there are no transaction I fail to see how can I ensure that my data > will not be corrupted? > Also since there is no constraints, only validation function, which are not > exactly the same because replication do not pass by them for clear reasons, > is that mean I have always asume the data is not valid? > > Thank you, > Ido > ------------------------------------------------------------------------ > VPRO www.vpro.nl > ------------------------------------------------------------------------ >