What is MRDI?
Multi-region delivery infrastructure (MRDI) is an architectural setup where the delivery infrastructure (servers, databases, routing, applications, etc.) is deployed in parallel with multiple geographical locations.
Each location has the capability to serve delivery traffic. If serving traffic from one location fails (due to an outage or another error), requests can be redirected to other locations, where they can be served successfully. We currently use two locations: the first location (primary region) serves the traffic by default and the other (secondary) region is only used in case the first one fails. We employ both automatic and manual failover to route traffic between regions.
The purpose of the MRDI is to ensure higher availability.
The naming comes from the AWS Regions on which we have built this functionality. It allows us to deploy our infrastructure in geographically distanced and physically independent locations so that an outage in one place (even a natural catastrophe) will not impact the service in the other place.
Basic API information
What are the business use cases of MRDI?
MRDI can be essential for the customers who are required to provide a multi-region infrastructure redundancy and higher availability SLAs to comply with internal or external compliance requirements.
In some cases, it may also be required to ensure the high availability of the content by law.
The MRDI helps in two use cases:
- When a content delivery request fails to be served by the primary region, the system automatically tries to serve the request from the secondary region.
- In the case of a primary region outage, the content delivery is taken over by secondary region delivery infrastructure to provide uninterrupted content delivery.
Will customers get better performance with the MRDI feature?
No, MRDI will not affect performance. MRDI is about high availability and compliance.
What APIs are multi-region with this feature?
Premium Plus customers who purchase the High Availability add-on (or former Enterprise plan customers with the High Availability platform) get access to MRDI.
To benefit from the MRDI customers need to use https://cdn-mr.contentful.com/ instead of https://cdn.contentful.com and https://graphql-mr.contentful.com instead of https://graphql.contentful.com to request their content.
The following APIs are available with MRDI:
- Content Delivery API (CDA).
- GraphQL Content API.
How does Contentful’s MRDI work with active-passive mode?
Contentful’s MRDI works in the active-passive (redundancy) mode. The main region is us-east-1 with live replicas in us-west-2. The failing requests in us-east-1 are retried in the us-west-2 region. In an event where the us-east-1 region fails, the failover would be automatically redirected to us-west-2.
The primary regions serve traffic under normal circumstances. The secondary region is used in case of an outage in the primary region.
What regions are used by Contentful?
Contentful is using three regions:
- Primary region: Virginia (us-east-1).
- Secondary region: Oregon (us-west-2).
- Each region uses 3 availability zones.
Are there European and Asian regions?
Currently, we do not provide multi-region availability covering the EMEA and APAC regions.
What content is covered by MRDI?
All published content is served through the Content Delivery (CDA) and GraphQL APIs.
What is the price for this feature?
This feature is only available to Premium Plus customers who purchase the High Availability add-on (or former Enterprise plan customers with the High Availability platform)
To learn more, contact your Account Executive.
How long does it take to set up for a customer?
The MRDI setup is not automatic. To start the process of initiating MRDI, you must contact an account executive or Customer Service Manager to notify the Infrastructure Group so they can complete the setup.
Contact your Account Executive to file a request to set up MRDI.
What about SLAs?
Contentful provides 99.99% availability SLA across the multi-region delivery infrastructure, which includes CDA and GraphQL.
The increased SLA only applies to the aforementioned APIs.