Optimal number of readers/writers Orthanc instances?

Google Groups archive
1

Hello! In my current Docker architecture, I have 4 Orthanc “reader” instances and 1 “writer” instance. The readers are load balanced by NGINX with the least_conn setting.

I was wondering if there is an optimal number of readers, and different strategies to a single Orthanc writer. Do you guys think it is possible to improve this 4-reader/1-writer architecture?

2

Hi Diego,

If you want to “improve” your setup, it means that you are observing “problems” right ? So I guess you should first describe your problems…

There’s no simple answer to load-balancing. It depends so much on

  • your storage (disk or object-storage)
  • your DICOM workload (are you storing, query/retrieving, moving ?)
  • your Rest API workload (same question)
  • the job engine usage
  • transcoding or not
  • …

Best regards,

Alain

3

Hi Allain. How are you doing? It was out of curiosity that I asked :grin:

Orthac actually works great for me. The only problem I have at the moment is that I am unable to delete very big exams without causing a bottleneck, therefore I am refactoring my architecture to contain a “router” Orthanc. By doing that I can switch the main Orthanc by a temporary one so that the main can focus on deleting exams for a certain time.

Having a router Orthanc makes it very easy to implement new writing strategies, like having one Orthanc for CT/MR (big exams) and another for CR/DX (small exams), or one Orthanc by institution for it to work on a completely isolated environment. Therefore I was wondering if someone else had any experience in these setups and is willing to share some information.

A quarta-feira, 13 de abril de 2022 à(s) 11:46:34 UTC+1, alain...@osimis.io escreveu: