Impact of Mass Relationship Creation on A12 Performance?

Data Services
1

Hi everyone,

We are currently developing a solution with a developer team to perform actions on a large number of documents in A12 at the same time—for example, booking many contracts at once. In this process, a large number of relationships are created between the ‘Contract’ document and the ‘BatchProcessRun’ document (potentially several thousand).

My question is: Does anyone have experience, to find out if creating such a high number of relationships at once could overwhelm A12’s capacity or seriously affect runtime/performance?

Thanks and best regards,

Timo

2

Hi Timo,
In our load test for A12 2024.06. We run a test that creates 180 CDDs/minute. This test was run with a database that already contained ~4.8 million documents.
The performance was measured:

  • Document creation time-consuming: 45 ~ 64ms/documents

CPU usage:

Memory usage: (the unit in here is byte, not KiB)

Note: In our test, the creation was done togethers with the other requests. It means that the system was not blocked during the creation process.

I hope this information is useful to you.

3

Hi @phuoc-sleek-sand,

can you provide us with additional data of the A12 application, resource settings and more importantly the cdd configuration used, e.g. was async enabled, and any useful setting.

thanks in advance.

4

Hi @ezekiel-risen-module ,
With these tests. The A12 application was configured with:

  • Client: 1 instance
  • Data service: 1 instance (With Asynchronous CDD mode was mention in documentation)
  • Artemis: 1 instance
  • CDD-JOB: 0 instance (we used 4 CDD-job instance to index the available data, but we shut them down when the initial indexing finished)
  • Hazel cast: enable
  • Solr: solr standalone
  • DB: External postgresql - 1 instance
  • Keycloak: 1 instance