Parallelization av import skriptet

Thomas Sødring tsodring at oslomet.no
Mon Aug 23 18:13:55 CEST 2021


Perhaps another thing worth considering is that nikita, as a spring-app, 
is multi-threaded. I think the default number of threads in 10 so if you 
parallelized the import code you might see an an increase in 
performance. It would also expose clearly if the database is a bottleneck.

Also, remember to not use the H2 database for this as you will likely 
see performance issues using an in-memory database.

  - Thomas

On 8/23/21 3:55 PM, Gareth Western wrote:
>
> I’m making some progress with testing the import of this large 
> uttrekk. The next issue I’d like to look at is performance. Serial 
> upload is too slow when the archive contains hundreds of thousands of 
> documents. A quick analysis of the structure indicates that there are 
> approximately 400,000 documents and 100,000 folders (mappe). Would it 
> be safe to do at this level? Or should I try at a higher level first 
> (e.g. klasse, of which there are approximately 4000)?
>
>
> _______________________________________________
> nikita-noark mailing list
> nikita-noark at nuug.no
> https://lists.nuug.no/mailman/listinfo/nikita-noark
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://lists.nuug.no/pipermail/nikita-noark/attachments/20210823/d3095527/attachment.htm 


More information about the nikita-noark mailing list