Multi-Cloud Architecture

Multi-cloud architecture uses services from more than one public cloud provider. It is a placement strategy, not a requirement to copy every workload everywhere.

The useful question is why a second provider belongs in the design. A provider may offer a required regional presence, a differentiated service, or a platform inherited through an acquisition. A separate provider may also host an independent workload. Each reason creates different architecture and operating requirements.

Google Cloud describes multi-cloud as a means to meet business and technical objectives, not a goal by itself. AWS makes the same point through workload placement guidance. Start with a named outcome. Then test whether another provider improves that outcome after you count the added complexity.

Choose a topology

There are three useful starting topologies.

Portfolio placement assigns separate workloads to separate providers. Each workload can use its provider's native services. Cross-cloud dependencies stay limited.

Partitioned application assigns distinct application components to different providers. This can satisfy a specific capability or placement constraint. It also creates cross-cloud data and network dependencies.