How to Design for Multi-Cloud Deployment - Best Practices and Considerations

Learn how to effectively design and implement multi-cloud deployments in this 40-minute CNCF conference talk. Explore key considerations including redundancy, availability, latency, geographic factors, and cost optimization. Discover strategies for selecting cloud providers, avoiding vendor lock-in, and planning architecture to meet specific requirements. Examine the benefits of hybrid cloud approaches, automation techniques, and global server load balancing. Gain insights into centralized monitoring, layer 4 security solutions, and real-world implementation examples from Snapt.

Intro
Redundancy & Availability
Latency, Geographic Considerations
Cost & Feature Options Best possible cloud for each workload Cloud can become platform
Forces design in a way that is not restrictive
Choosing cloud providers
Don't design for a specific cloud Lock-in dangers as mentioned Platform restrictions -e.g., Kubernetes
Plan your architecture Locations requirements, etc.
Ensure your requirements are met Databases Egress data fees / structure Staff abilities and training Regulatory requirements
Investigate commercials Contracts and commercials Support and SLA Reliability and performance guarantees
Consider 2x clouds or 1 cloud + X Hybrid cloud, hybrid platforms Custom datacenters App-based services and deployments
Consider your architecture
Try to automate deployments
Utilize auto-scaling functions Don't pay for 100% high availability Ensure AWS outage causes scaling in GCP, etc.
Route traffic with GSLB Global Server Load Balancing Intelligent DNS server Allows for test deployments, weighting
Centralized Monitoring
3. Cloud For Layer 4 Solutions Handle layer 4 security at the cloud level Security groups, etc.
How Snapt does it?