How much actually carries over when you run across multiple clouds?

November 21, 2026 5 answers
Asked by _b57~mivdelo.8

I'm looking at the practical trade-offs of using AWS and Google Cloud together. How much transfers between them, and what gets more complicated for networking, Kubernetes, Terraform, and CI/CD?

Answers

3m~canvaspulley_617

Start with the reason for using two clouds. If the goal is resilience, multiple regions in one provider may be simpler. If the goal is reducing vendor lock-in, that can be worth pursuing, but it comes with extra governance, operational work, and potentially less leverage for spending discounts with one provider.

It does not have to mean duplicating every workload. The useful question is which applications and dependencies actually need to run in both places.

Reply to this answer
c.418!zunvirox

The concepts carry over better than the configurations. Networking and identity are good examples. AWS VPCs are regional with subnets tied to availability zones. Google Cloud VPCs are global with regional subnets. That difference affects IP planning and how you think about connectivity.

Identity is another trap. AWS roles and Google Cloud service accounts can look analogous, but don't assume they behave the same way. Review permissions, trust relationships, and the identity model separately rather than translating names one for one.

Reply to this answer
q.631!pluvnori

Put the data flows on a diagram before choosing an interconnect. A site-to-site VPN is one option, while dedicated connectivity adds another set of design and operational considerations. Check which regions need to talk and whether traffic has to cross regions within either provider.

Price out egress between clouds and cross-region traffic before committing. Data tends to anchor applications, so moving lots of it back and forth can make an otherwise neat architecture expensive.

_b57~mivdelo.8:

So a dedicated connection doesn't necessarily remove the need to plan for cross-region traffic?

3e!ferrulevisor_615:

Right. It depends on where the connection lands and where the workloads run. Map the full path and its costs, not just the link between providers.

Reply to this answer
_c48+brinvelo.7

Kubernetes gives you a common way to run applications, but it doesn't make the clouds interchangeable. Networking, identity, and integrations with managed services still need provider-specific work. Custom resources can help connect those pieces, though they also become things you have to maintain.

I'd use it where a shared application platform is useful, not just because there are two cloud accounts.

Reply to this answer
Nuvdavi-3k

Terraform and CI/CD can give both environments a common workflow, but the underlying provider configurations will still differ. Keep those differences visible rather than assuming one configuration maps neatly to both. Budget alerts and consistent tagging can also help you keep track of spending across the two.

Reply to this answer

Add your answer

Answers are reviewed before publication.

Related Discussions