Guidance on Scaling Solutions for High-Traffic Sites

[Jakub Borkowski]

Hi,

I’m seeking advice on best practices for scaling solutions on websites with a large user base. Are there any specific recommendations or guidelines you’d suggest?

Additionally, could you share your insights on which platform might be more effective in this context: App Engine or Cloud Run?

Thank you in advance for your guidance!

Best regards,
Jakub

[Instant BQML and Vertex Users]

Hi Jakub,

Thanks for your questions about scaling and platform choice for large user bases.

Scaling:

• CRMint is designed to handle Measurement Protocol integration for GA4 properties of any size. You generally shouldn't encounter scaling issues related to data volume.
• Measurement Protocol Batch Size: You can experiment with increasing the batch size in the "Events to GA4" and "Percentiles to GA4" jobs within your Instant BQML/Vertex prediction pipelines. However, the standard batch size should be sufficient for most cases.

App Engine vs. Cloud Run:

• Both platforms are capable of running CRMint effectively. The choice depends mainly on your preferences and existing infrastructure.
• App Engine:
  • Simpler to troubleshoot (in my experience).
  • Might be preferable if you're already familiar with App Engine.
• Cloud Run:
  • Requires a Google Cloud Organization (through Workspace or Cloud Identity) in your project.
  • Offers more flexibility and control over the deployment environment.

Recommendation:

• If you're unsure about setting up a Google Cloud Organization or haven't done so already, I recommend sticking with App Engine for your CRMint deployment. It's generally easier to get started with and offers excellent performance for most use cases.

Ultimately, the best platform choice depends on your specific needs and familiarity with Google Cloud services. Feel free to reach out if you have any further questions or would like to discuss your specific requirements in more detail.