Re:OrientDB: A new Open Source NoSQL DBMS
by alexpopescu on Wednesday May 26, 2010 @04:24AM
(#1379810)
Other option may be BerkleyDB/Java. It should be pretty easy to get started. In terms of querying things may be a bit more difficult (as for any KV store), but as far as I remember there are ways to programmatically define secondary indexes. (note: I'm not affiliated with Oracle/Sun/Berkley/etc. :-)
Have a look at Riak. It scales up (and down) very well. And it's been in production use for a while now.
Thanks. I've had a look and from what I can see insertion is by HTTP or JSON only, and not over persistent connection. I expect this will be too slow for the volume of data I have.
I've got an app in mind, and have some experience with couchdb but not with the systems you mention.
Orient interested me because I'm constantly on the lookout for something with these features:
* Key/value database.
* Straightforward to install on linux and darwin.
* Able to push data into it through high-bandwidth
* Queries give result sets that are reliably correct for a moment in time.
* Doesn't take up large amounts of disk space.
* Straightforward to get started with and evaluate without large time investment.
Couch fails a few of them, but I didn't know they were requirements until I'd started playing with it.
Orient seems to be very easy to set up, and supports a binary input protocol which might be faster than the HTTP drop mechanism I've been using against couchdb.