APIM Premium with multiple gateway deployments

Katare, Ashish 141 Reputation points

In a recent call with Microsoft it came to my notice the way APIM premium is being priced -

Use case - one APIM Premium instance with multiple gateways based on regions/location - i would charged for number of gateways multiplied by the amount for one APIM instance.
so if i have 3 gateways in different regions under one APIM instance then i would be charged close to 3*$2,795.17/month (which is a lot I believe)


Trying to see the options to reduce the cost or if can use APIM Basic along with Multi region support for our API Management program..

Appreciate all the inputs around reducing the cost in Azure APIM

Azure API Management
Azure API Management
An Azure service that provides a hybrid, multi-cloud management platform for APIs.
1,951 questions
{count} votes

Accepted answer
  1. Pramod Valavala 20,611 Reputation points Microsoft Employee

    @Katare, Ashish Since multi region deployments are supported only the Premium Tier, you could try and mimic the same deployment strategy but with basic/standard instances and without a single control instance.

    And the best way to achieve this would be to

    1. Setup CI/CD with ARM Templates to ensure the same configuration is deployed to each region
    2. Setup Traffic Manager for routing

    But do note that the Premium Tier instances can support many more requests compared to basic/standard tiers. So, depending on your load, you might need multiple smaller instances to handle the same amount of load, which may come up to the same cost with extra work to synchronize all instances.

    Another option you could consider would be to leverage Self Hosted Gateways that you could host in your VNET directly. If the services in the VNET are hosted as containers in AKS for example, the self-hosted gateway could be hosted there as well. Otherwise, you could leverage Azure Container Instances to run the gateway as well.

    Note that this still requires APIM Premium Tier would with just one fully managed instance and depending on your projected load, you would allocate compute resources accordingly for the self-hosted gateway containers wherever they are hosted.

    0 comments No comments

0 additional answers

Sort by: Most helpful