You are viewing a plain text version of this content. The canonical link for it is here.

Posted to users@jena.apache.org by "Dimov, Stefan" <st...@sap.com> on 2018/01/05 06:15:47 UTC

Upload into in-mem DB

What is the fastest way to upload multiple RDF (turtle) files into in-memory TDB?

Regards,
Stefan

Re: Upload into in-mem DB

Posted by Andy Seaborne <an...@apache.org>.

In-memory TDB is slow.

It get sit "in-memory" but having a fake disk where it reads and writes 
"disk" blocks.  It is great for testing - no real I/O overhead to create 
a small database - and it exercises all the TDB machinery except for the 
lowest level of file handling.

Jena does offer TIM - "Transactions In Memory" - 
Dataset[Graph]Factory.createTxnMem.

It is much faster than in-memory TDB.
It transactions are the same semantics as TDB.

It loads a bit faster than than parser can parse Turtle (Turtle is 
slower than N-Triples even on the same NT-data - the parser inner loop 
is more complex, less CPU cache benefits).

     Andy

On 05/01/18 06:15, Dimov, Stefan wrote:
> What is the fastest way to upload multiple RDF (turtle) files into in-memory TDB?
> 
> Regards,
> Stefan
>