コンテンツにスキップ

Performance Tuning

Depending on the monitored environment, it may be necessary to tune the resources of RISKEN servers.

RISKEN adopts a microservices architecture, which allows fine-tuning of compute resources for each service.


Preparation

  • Make sure that there are sufficient scalable compute resources on the Node side in advance.

Scaling Out

If there are many users or projects and the overall processing time is taking too long, consider performing Scaling Out of the target service.

  • By adjusting the replicas of the Deployment, it is possible to increase the number of simultaneous executions in parallel processing.
  • For details on how to configure this, please refer to the official Kubernetes documentation .

Spec Tuning

If a scan of a large data source is required and resources such as CPU and MEM are running out, consider performing Scaling Up of the target service.

  • By adjusting the resources of the Deployment, it is possible to increase the compute resources.
  • For details on how to tune the specs, please refer to the official Kubernetes documentation .
    • Determining the optimal request/limit for memory or CPU resources requires measuring while actually running.
    • As an example of tuning ideas and measurement methods, we introduce a blog by Sysdig .
  • RISKEN's individual services mainly consume memory or CPU resources, but some scanners temporarily write data to disk (such as the data to be scanned or the scan result files).
    • Therefore, it is unnecessary to add persistent volumes to the containers running scans, but ephemeral volumes may need to be added.