GCP Integration with Dynatrace

[Dg03cloud]

I would like to integrate GCP with Dynatrace monitoring service to analyze the logs and metrics from gcp. For this usecase, I think to use a GKE auto pilot cluster to forward the logs from gcp to dynatrace. Is there a helpful references with example on this usecase would be helpful?

[ms4446]

I suggest reaching out to Dynatrace for customer use cases and referrals. For your reference here are some helpful links I found:

• Dynatrace Setup and Configuration

  • Refer to Dynatrace's documentation on Google Cloud Platform integrations: https://docs.dynatrace.com/docs/setup-and-configuration/setup-on-cloud-platforms/google-cloud-platfo...

• GKE Autopilot Cluster Creation

  • Follow the GCP guide for creating a GKE Autopilot cluster: https://cloud.google.com/kubernetes-engine/docs/how-to/creating-an-autopilot-cluster

• Dynatrace OneAgent Installation on GKE

  • Dynatrace provides instructions for deploying OneAgent on GKE: https://docs.dynatrace.com/docs/setup-and-configuration/setup-on-cloud-platforms/google-cloud-platfo...

• GCP Metric and Log Integration

  • This is the crucial part for forwarding logs to Dynatrace. Follow Dynatrace's detailed guide: https://docs.dynatrace.com/docs/setup-and-configuration/setup-on-cloud-platforms/google-cloud-platfo...

[Dg03cloud]

Thanks for your prompt reply. As per the Dynatrace documentation, it’s recommended to use a GKE auto pilot cluster (with Linux image). But based on which factors, do we need to make a call to use a GKE auto pilot or a GKE standard cluster for integration with Dynatrace ? Any thoughts on this apart of cost efficiency?

[ms4446 · Accepted solution]

Here some considerations that influence the choice between GKE Autopilot and GKE Standard for Dynatrace integration:

• Management and Control:

  • GKE Autopilot: Google manages the underlying infrastructure (nodes, etc.), handles node-level operations, scaling, and upgrades. You have less control over the low-level details of your cluster.
  • GKE Standard: You get full administrative control over the cluster's nodes, allowing extensive customization of the Kubernetes environment.

• Operational Simplicity:

  • GKE Autopilot: Ideal if you want the easiest setup and maintenance. Autopilot automates tasks, making it a great choice if your team doesn't need in-depth Kubernetes expertise.
  • GKE Standard: Requires your team to have stronger Kubernetes knowledge for management and troubleshooting.

• Workload Requirements:

  • GKE Autopilot: May have limitations depending on specific workload needs. Thoroughly assess if Autopilot supports all features required by your applications.
  • GKE Standard: Provides greater flexibility to tailor the cluster to any workload due to granular control.

• Customization:

  • GKE Autopilot: Limited customization options as Google manages much of the infrastructure.
  • GKE Standard: Full flexibility to customize node configurations, network settings, etc.

• Cost Efficiency:

  • GKE Autopilot: Pay-per-pod pricing model, potentially more cost-effective for well-defined workloads.
  • GKE Standard: Cost calculated based on node resources. Consider this if you need predictable costs or have workloads sensitive to node-level changes.

When GKE Autopilot might be a good fit:

• You prioritize operational simplicity and managed infrastructure.
• Your team focuses on application development rather than deep Kubernetes management.
• Your workloads fit well within GKE Autopilot's capabilities.
• You can benefit from the pay-per-pod cost structure.

When GKE Standard might be more suitable:

• You require fine-grained control over nodes, networking, or the Kubernetes environment.
• You have specialized workloads that need a highly customized cluster setup.
• Your team has strong Kubernetes expertise and can handle advanced cluster management.
• You prefer the predictable cost structure based on node resources.

Beyond Cost Efficiency:

While cost is important, the decision depends heavily on your team's skillset, your management preferences, and your workload's specific requirements.

If you're unsure, start with a GKE Autopilot cluster for its simplicity. If you later encounter limitations or the need for advanced customization, a transition to GKE Standard is possible.