I'm currently considering migrating part of my project to public clouds and would like to gather general recommendations on service selection, building a scalable architecture, and managing the budget. I'm particularly interested in what approaches to security and deployment automation are considered best practices, and how to avoid common pitfalls when working with multiple providers simultaneously. Does anyone have proven methodologies or links to useful resources? It would be great to discuss experiences and compile a list of questions for further study.
What are the best practices for selecting services and tools when migrating to public clouds?
👁️ 88 views💬 2 replies❤️ 0 likes
2 Replies
When choosing public cloud services, I immediately focus on the "versatility + cost" factor. For basic needs, a combination of IaaS (VMs + network rules) and PaaS (Managed Databases, Message Queues) often suffices. I always conduct a matrix-based comparative analysis of cost/performance and supported SLIs to identify where I can save without compromising fault tolerance. For budget management, tag-based labels and automated alerting tools like AWS Budgets or Azure Cost Management API, integrated with Terraform/Pulumi for expense tracking in CI/CD pipelines, are incredibly helpful.
As for security and automation, I believe the best practice is "Infrastructure as Code + Policy as Code." I store all infrastructure descriptions in Git and layer OPA/Terraform Sentinel policies on top to enforce IAM permissions, encryption, and data placement. When working with multiple providers, I use abstractions through Crossplane or Terraform provider modules to avoid vendor lock-in and simplify portability. I also implement multi-level test playthroughs (using LocalStack/Azurite) and write smoke tests for deployments to catch any early-stage pitfalls. Helpful resources include the AWS Well-Architected Framework, Google Cloud Architecture Center, Microsoft Cloud Adoption Framework, and the book "Designing Data-Intensive Applications."
Start by selecting a single provider and describe the infrastructure as code (Terraform/Pulumi), so you can easily switch between providers and automate deployments via CI/CD. Use resource tags and budget limits, and implement security through IAM policies and network zones, validating them at every migration stage. This approach has already helped me avoid access-right conflicts and unexpected cost spikes when working across multiple clouds.