AWS, Azure & GCP · Cloud across providers
What is hybrid cloud and how do you connect it?
Short answer
Hybrid cloud runs workloads across on-premises and cloud with shared identity, networking and operations. It is driven by data residency, latency, legacy systems or gradual migration, and needs reliable private connectivity.
Hybrid is not just a VPN between two environments. It is an operating model where identity, networking, observability and deployment span both sides. Without that integration, you get two disconnected estates and a lot of manual work.
Connectivity options
- Site-to-site VPN: quick to set up, encrypted, but dependent on internet quality
- Dedicated interconnect: private bandwidth such as Direct Connect, ExpressRoute or Cloud Interconnect
- Private endpoints and private link for service-level access without public exposure
- Identity federation so users and workloads authenticate consistently
- DNS forwarding and consistent naming across environments
Latency and data gravity shape the design. Large datasets are expensive to move repeatedly, so processing often stays near the data. Egress charges and backup traffic also need to be modelled before committing to a split.
Operational considerations
- A single identity and access model across both environments
- Consistent logging, tracing and incident response
- A deployment pipeline that can target both sides safely
- Clear ownership for network links and failover behaviour
Follow-ups you should expect
Interviewers use these to check whether the first answer was understanding or memorisation.
- 1When is a VPN sufficient versus a dedicated interconnect?
- 2How does data gravity affect hybrid design?
- 3How do you keep identity consistent across environments?
- 4What happens when the private link fails?
Say it in the interview
Hybrid cloud means workloads run across on-premises and cloud with shared identity, networking and operations, not just a VPN. I choose connectivity based on bandwidth, latency and criticality: VPN for quick or lower-volume links, dedicated interconnect for production. I also plan for link failure and data gravity, because moving large datasets repeatedly is slow and expensive.
Use this as a model, not a script. Saying it in your own words is what makes it survive follow-up pressure.
Related questions
Reading is step one
Now say it out loud.
Understanding an answer and delivering it under pressure are different skills. Rehearse this topic in a live AWS, Azure & GCP mock interview and get scored feedback on depth, clarity and confidence.