Sarven Capadisli (Commented) (JIRA) wrote:
> Given my preferences, do you reckon that tdbloader2 is more suitable?

Hi Sarven,
I use tdbloader2 myself and my advise is alone the lines of what others have 
said.

I use a 64 bit (Linux) machine with enough RAM (to fit the node table in it).
If I have RDF data not in N-Triple|N-Quad format, I convert (and validate) it 
into N-Triple|N-Quad format.
I find it easier to deal with a large and compressed .nt|.nq file, rather than 
a lot of small files (this is what Hadoop also likes, uncompressed in that 
case).

With enough RAM you should not have problems loading 500 million triples into 
an empty DB.

Paolo

Reply via email to