Multi-Cloud Deployment Paradigm
Multi-cloud Atlas deployments set up cluster nodes across multiple geographic regions and multiple cloud providers. Multi-cloud deployments enhance protection in the case of a regional outage or cloud provider outage by automatically rerouting traffic to a node in another region for continuous availability and a smooth user experience. Multi-cloud deployments can also protect against vendor lock-in, enhance performance, and help meet compliance requirements for data sovereignty.
Atlas supports multi-cloud deployment across any combination of AWS, Azure, and Google Cloud.
To learn how to configure multi-cloud deployments and learn about the different types of nodes you can add, see Configure High Availability and Workload Isolation in the Atlas documentation.
Multi-region deployments can deploy nodes across a single cloud provider or multiple cloud providers. As each cloud provider has its own set of regions, multi-cloud deployments are also multi-region deployments. To learn about single-cloud multi-region deployments, see Single-Region Deployment Paradigm.
To learn about global deployments, which support zone configuration across multiple cloud providers, see Global Deployment Paradigm.
The following diagram shows a multi-region, multi-cloud Atlas deployment for regions that support availability zones:
Use Cases for Multi-Cloud Deployments
To learn the use cases for multi-region, single-cloud deployments, see Single-Region Deployment Paradigm.
Multi-region, single-cloud deployments are best for the following use cases:
Region-Specific Applications that Require Low Latency and High Availability (Multiple Cloud Providers)
To improve performance for critical operations, all Atlas deployments allow you to deploy data close to your users' geographic location, which reduces latency. However, deploying to a single region or single cloud provider sacrifices high availability if there is a regional or cloud provider outage. Additionally, deploying to a single cloud provider can make it difficult to change vendors later in case of pricing changes or other cloud provider changes. You can configure a multi-region, multi-cloud deployment for high availability, low latency, and avoid vendor lock-in.
A multi-region, single-cloud deployment may be best for you if you have the following requirements:
You want to use multiple cloud providers
You want to deploy to more than one region for high availability
Your application requires low latency and has a majority of users in one geographic location, since multi-region deployments allow you to deploy nodes to different regions within the same geographic area.
For example, for an application deployed with AWS and Azure with
users primarily located in the US, you can deploy a multi-region,
multi-cloud deployment to three regions within the US (such as
us-east-1 and us-east-2 for AWS, and eastus for Azure).
This ensures low latency since all regions are within the eastern US,
while offering high availability if there's a regional outage or a cloud
provider outage that affects the primary
node. Additionally, it makes it easier to transition off of one of these
cloud providers if needed.
If your application requires low latency for users in any region across a global user base, consider a Global Deployment Paradigm.
If your application requires low latency but doesn't require cross-region or cross-provider high availability, consider a Single-Region Deployment Paradigm.
If your application requires low latency, high availability, and deployment across a single cloud provider, consider a Multi-Region Deployment Paradigm.
Applications that Require Data Sovereignty and High Availability (Multiple Cloud Providers)
For compliance with data residency laws, data can be partitioned to reside in specific regions, ensuring adherence to local regulations. However, deploying to a single region or single cloud provider sacrifices high availability if there is a regional or cloud provider outage. Additionally, deploying to a single cloud provider can make it difficult to change vendors later in case of pricing changes or other cloud provider changes. You can configure a multi-region, multi-cloud deployment for high availability, data sovereignty, and avoid vendor lock-in.
A multi-region, multi-cloud deployment may be best for you if you have the following requirements:
You want to use multiple cloud providers
You want to deploy to more than one region for high availability
Your application requires data sovereignty, since multi-region deployments allow you to deploy nodes to different regions within the same geographic area.
For example, for an application deployed with AWS and Azure that
requires data storage in Europe, you can deploy a multi-region,
multi-cloud deployment to three regions within the EU (such as
eu-west-1 and eu-west-2 for AWS, and uksouth for Azure).
This ensures data sovereignty since all regions are within the EU, while
offering high availability if there's a regional outage or a cloud
provider outage that affects the primary node. Additionally, it makes
it easier to transition off of one of these cloud providers if needed.
If your application requires data sovereignty but doesn't require cross-region or cross-provider high availability, consider a Single-Region Deployment Paradigm.
If your application requires data sovereignty and deployment across a single cloud provider, consider a Multi-Region Deployment Paradigm.
Considerations for Multi-Cloud Deployments
Other considerations for multi-region deployments include:
High availability depends on the deployment of nodes across regions as well as the number, distribution, and priority order of nodes. To learn more about recommended cluster topologies for high availability, see Guidance for Atlas High Availability.
Multi-cloud deployments are available only for
M10dedicated clusters and larger.
For more considerations, see Considerations in the Atlas documentation.
Recommendations for Multi-Cloud Deployments
To learn about recommendations that apply to multi-region and multi-cloud deployments, see the following sections:
Important
All recommendations that apply to multi-region deployments also apply to multi-cloud deployments because all multi-cloud deployments are also multi-region. There may be other considerations for multi-cloud deployments that are not covered in the Atlas Architecture Center, such as cloud-provider-aware application-side settings. Contact MongoDB's Professional Services team to create a custom landing zone for your Atlas multi-cloud deployments that covers all considerations.