How much resources are used in Data Services?

Peace

we’re asking ourselves how much RAM and CPU are used by an unmodified Data Services application. Does someone have already measured numbers ?

You can assume, that there is only one small DM, FM, OM and AM in case this is necessary.
Maybe someone can tell how these models have impact on the resources.

Best regards,
Tjorben

Hello @tjorben-atomic-moss

Thanks for interesting question. I think Load Test team could give us such numbers. I think Project Template based application with 1DM, FM, OM and AM could be good source.

@phuoc-sleek-sand can you please have a look?

Hi @tjorben-atomic-moss ,
I can do a quick mearsure for the resources usage. Which DS version that you need to use?
It will be better if you can provide the infrastructure that you want to use in your application.

  • solr: embedded/external standalone/solr cloud
  • DB: embedded/external

Hi @tjorben-atomic-moss,

until you specify the measurement conditions with @phuoc-sleek-sand, our currently-ruuning Project Template instance on cluster (with 1 DM, FM, OM and AM) consumed max 681MB of RAM:
image
and max 0.0218 of CPU:
image
but CPU is hard to quantify generally.

Hey and thanks for your replies.

To you @phuoc-sleek-sand:

We would use an external DB and an external stnadalone solr, but it would be nice to know how much resources are used there too, because we need both together afaik, so the resource usage of the solr instance has to be added

Hi @tjorben-atomic-moss ,
I tested deployment a simple app base on A12 project template with 1 DM, FM, OM and AM.
The infrastructure:

  • Data-service: 1 instance
  • Client: 1 instance
  • Postgresql: 1 instance
  • Solr: 1 instance
  • keycloak: 1 instance

Here is the detail:
Memory: The highest memory usage is solr with ~1.25 Gi, DS is 720M, keycloak is 515M and postgresql is 80M

CPU: This is quite hare to have a general number. It depend on how much process will be run in your application. The chart below show the CPU usage from statup until all services is up.

Thank you very much !

Do you know roughly how the resource usage increase with 2 DM, 2 FM and 2 OM ?
It might be important to know how the resource usage scales with new models

Best regards,
Tjorben

Edit: Do you know if i can use the same solr instance for multiple A12 instances (Data Service applications) ?

Hi @tjorben-atomic-moss ,
Increasing a small number of model will not affect much to the resource usage. The resouce usage will be depended on how much data in your DB and how many user do CRUD at the same time.
Currently we have deployment template with these default configuration:

  • DS:
  limits:
    memory: 4096Mi
    cpu: 2
  requests:
    memory: 2048Mi
    cpu: 0.5
  • Client:
  limits:
    memory: 256Mi
    cpu: 0.5
  requests:
    memory: 128Mi
    cpu: 0.1
  • Postgresql:
      requests:
        memory: 256Mi
        cpu: 100m
  • Solr:
    limits:
      cpu: 2
      memory: 3Gi
    requests:
      cpu: 0.5
      memory: 2Gi
  • Keycloak:
  requests:
    cpu: 2
    memory: 512Mi
  limits:
    cpu: 3
    memory: 1024Mi

These configuration is stable for a simple A12 base application. If you want to have scalability testing for your application, you can contact to mgm load test team.

Do you know if i can use the same solr instance for multiple A12 instances (Data Service applications) ?
=> yes, you can. You just need to aware that indexing multiple app data in to 1 solr may effect to your application performance. And it also increase the resource usage in solr.

Ty for you answers :slight_smile: