Declining TDB load performance with larger files

I'm loading a 111 million triples file (GND German Authority files). For the first roughly 70 million triples, it's really fast (more than 60,000 avg), but then throughput declines continuously to a thousand or just some hundred triples (which brings down the avg to less than 7000). During the

Re: Declining TDB load performance with larger files

Hi Joachim, What is happing is that the system is running out of working space and the disk is being used for real. JAVA_OPTS: -d64 -Xms6g -Xmx10g Don't set -Xmx10g. Try a 2G heap. Don't bother with -Xms. More heap does not help - in fact, it can make it worse. TDB uses memory mapped

AW: Declining TDB load performance with larger files

Hi Andy, Thank you so much. I fixed the heap issue, and will check and report my findings. Cheers, Joachim -Ursprüngliche Nachricht- Von: Andy Seaborne [mailto:a...@apache.org] Gesendet: Montag, 28. Oktober 2013 16:58 An: users@jena.apache.org Betreff: Re: Declining TDB load

Re: jena-text assembler file not working for me

Thanks Andy, I didn't think to look at that, because apparently I misunderstood unionDefaultGraph. I thought it was analagous to Allegrograph which treats the default graph as a union of all graphs (default + named). According to: 

Re: jena-text assembler file not working for me

On 28/10/13 18:21, Tim Harsch wrote: Thanks Andy, I didn't think to look at that, because apparently I misunderstood unionDefaultGraph. I thought it was analagous to Allegrograph which treats the default graph as a union of all graphs (default + named). According to: