Data Stores vs. ODBMSs
I asked Anat Gafni, VP of Engineering at db4objects, on her opinion on how ODBMSs compare with respect to new “data stores”, such as “document stores”, and “nosql databases”.
RVZ: Anat, systems such as CouchDB, MongoDB, SimpleDB, Voldemort, Scalaris, etc. provide less functionality than OODBs but a distributed “object” cache over multiple machines. How do they compare with respect to ODBMSs?
Anat: We can categorize “stores” and see if they are more similar or different than OODBs, in a couple of ways:
By each dimension of the purpose of OODBS:
1. persistent (could be accomplished by other methods like: replicating to other machines, using non-volatile caches, etc.)
2. Being queriable
3. Scalable (beyond what can be in cache, but could be distributed instead)
4. Objects vs.Relations
5. can express and query based on complex relationships among data items
6. can be shared among multiple “clients”
Many of these other database are similar to oodbs in item 1, 3 and 4. I am not sure they have capabilities in 3, 5 and 6 above.
There is a lot of interest in the USA in particular, w.r.t. to internet based applications and cloud computing. Big_table, and such look to me more like an algorithm based on traditional stores, rather than a db.
Anat Gafni has over 20 years of experience in managing software development and product strategy. At db4objects she is responsible for managing engineering and support. Anat earned her Ph.D. in Computer Science from the University of Southern California, and a MA degree from Boston University in Math and Computer Science.