Well why not using the Java API or the ETL tool to import huge amounts of 
data?

my inserts of raw data is by ~10k/sec on bad hardware. Or is it needed to 
do it with nodejs?

Am Dienstag, 23. September 2014 05:07:42 UTC+2 schrieb Luke Rossy:
>
> Hi, I am new to OrientDB and I have some questions regarding write 
> performance. 
>
> We used a Document based DB, and then tried to dump a sample set of data 
> and start playing with it to learn more about OrientDB
>
> We started with the 1.7.9 release and set it up on two servers with the 
> following specs.
>
> Debian 7
> 8GB DDR3-ECC Un-Buffered RAM
> 1x 1TB HDD
> Xeon E31230 V2 (3.30GHz)
> Java 7
>
> Then, on another server on the same LAN, we had a nodejs script read from 
> a flat file, and used the nodejs client (
> https://github.com/codemix/oriento 
> <https://www.google.com/url?q=https%3A%2F%2Fgithub.com%2Fcodemix%2Foriento&sa=D&sntz=1&usg=AFQjCNEHi0Don9lTDapdaQjGpasXYa3ePQ>)
>  
> to insert the data row by row into OrientDB. 
>
> Our script was maxed out at ~350 inserts a second. 
>
> Then, we installed 2.0-M1 and did the exact same test, and the write 
> performance went up, but only to 600 inserts / second
>
> Finally, I killed the one of the two nodes, and then the performance rose 
> to 3000 inserts/sec. 
>
> Are there any configuration options I should be tuning for write heavy 
> databases? 
>  
> I have already tried Dstorage.diskCache.bufferSize but there was no 
> significant change. 
> I have also changed writeQuorum = 1 (even though that was against your 
> recommendation in the docs) 
>
> Why is there such a massive difference between distributed and stand alone 
> performance? What can I do to minimize this?
>
>
> Thank you 
>
>
>
>

-- 

--- 
You received this message because you are subscribed to the Google Groups 
"OrientDB" 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