I am less and less convinced that we should ever rely on a single database 
as "the best candidate" for persistence - this is the Oracle or SQLServer 
discussion based on the misguided notion that simplification comes from 
consolidating data in to a single vendor product.  What differentiates 
database technologies is the CAP and index properties they use to let you 
write and retrieve data. A graph index is different to a columnar to a set, 
to a vector etc.

Best always depends on the usecase. While that sounds like a copout, what 
it means to say is that information access requirements should be informing 
your choice of databases. Put another way, you might store some bulk data 
in a columnar store and meta-data in a graph and some content in a KV 
store. Polyglot persistence is the trend in this space.

Specifically - I wouldn't be using Neo4j to write device based event 
streams where Cassandra or HDFS is better optimised for that kind of high 
volume write scenario. However, I might calculate and store meta data 
findings from those streams in to Neo4j. 


hth

-- 
You received this message because you are subscribed to the Google Groups 
"Neo4j" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to