Hi Andy,
Please find my comments.
---------- Forwarded message ----------
From: Andy Seaborne <[email protected]>
Date: 2014-01-28
Subject: Re: Fwd: Spaql queries optimizations for Freebase set
To: [email protected]
On 28/01/14 10:54, Ewa Szwed wrote:
> Hi,
> Thanks so much for this answer.
> I am definitely going to test the changing order of filter/binding
> constructs.
> For my system configuration I run Jena and Fuseki on one virtual machine
> with Linux Red Hat distribution and of 8 cores CPU, 64 GB RAM and 1.2 TB
> hard drive.
> I have configured Fuseki java -Xmx to 12000M.
>
Too large for TDB unless you are doing large update transactions.
Most of the memory used is out-of-heap.
Initially I had it set at 1200M and I saw OutOfMemory errors for some
queries.
The response size for my queries vary and can be from 600KB for 600MB.
I can reduce to 6000MB, would that be recommended.
My jena data directory size is 175 GB
> I am storing more than 2 000 000 000 triples (the number of facts in
> Freebase).
> Ewa
>
Ewa,
Could you try an experiment for me please?
Could you see if changing the "swapiness" makes a difference?
Normally it's 60 but if you could try, say, 40, 20, 10, 0. There have been
reports of odd effects reported but no reproducible test cases.
I will check that and come back to you.
Andy
[1]
https://access.redhat.com/site/documentation/en-US/Red_
Hat_Directory_Server/8.2/html/Performance_Tuning_Guide/system-tuning.html
http://askubuntu.com/questions/103915/how-do-i-configure-swappiness
