2 Answers
- Newest
- Most votes
- Most comments
1
Answer is it depends :)
It depends on compliance, regulatory and company-specific requirements.
There are typical two models I have seen in my experience:
- A centralized approach, where another ingress VPC contains an API Gateway (AWS or 3th party as Kong, Axway, etc..) or Application Load Balancer. This is the central place for governance and controlling all internal APIs who want to become public and should handle DooS, WAF, etc.. Inspection may be needed: https://aws.amazon.com/blogs/networking-and-content-delivery/centralized-inspection-architecture-with-aws-gateway-load-balancer-and-aws-transit-gateway/. This entails a potential bottleneckness as central "governer" is responsible.
- A distributed approach where individual "Spoke" VPCs are allowed to expose APIs publicly. This may be a bit more complex but will probably ensure more team independence.
In both cases IaC and clear responsibilities are key.
1
Hi, to get a broad perspective on all what is possible, I'd suggest you to read this very comprehensive wp detailling all what's possible:
Hope it helps!
Didier
Relevant content
- asked 4 months ago
- AWS OFFICIALUpdated 2 months ago
- AWS OFFICIALUpdated 4 years ago
