In 2020, Microsoft has set ambitious sustainability goals to be carbon negative by 2030, remove historical carbon emissions by 2050 and more. We’ve made significant progress in the past years, but Microsoft is not the only company with ambitious goals.

To help our customers, we’ve launched Azure Carbon Optimization to help measuring carbon emissions and optimizations in Azure for their workloads, offer guidance related to sustainable workloads, Microsoft Sustainability Manager to gain more insights and more.

Customers based in Europe can leverage these reports to comply with the Corporate Sustainability Reporting Directive (CSRD) as they are required to report their carbon emissions.

But why stop there?

We are pleased to announce the public preview of environmental sustainability features in Azure API Management designed to help minimize the environmental impact of your API infrastructure.

API Management allows you to influence your API landscape based on Scope 2 emissions:

• Shift and load-balance API traffic to backend regions with lower carbon intensity
• Shape API traffic based on carbon emissions in your API Management service's region

Let’s dive in and see how we can help.

Shifting your traffic to greener regions

As API customers direct traffic to your backends, we can ensure the routing is exclusively managed to those operating within green regions.

Load balanced pools can now become carbon-aware, automatically shifting API load from high-carbon to greener Azure regions based on your configured backends, helping reduce environmental impact.

To get started, you can easily attribute the Azure region in which your backend is running:

{
    "type": "Microsoft.ApiManagement/service/backends", 
    "apiVersion": "2024-10-01-preview", 
    "name": "sustainable-backend", 
    "properties": {
        "url": "https://mybackend.example.com",
        "protocol": "http",
        "azureRegion": "westeurope",
        [...]
  }
}

Once configuration is complete, you may update the load balancer backend to specify a preferred carbon emission level for your regionalized backends.

This approach ensures traffic is routed only to backends that meet your company sustainability criteria for acceptable emissions in each region. You can select from one of the five emission categories we have introduced.

{
    [...]
    "properties": {
        "description": "Load balancer for multiple backends",
        "type": "Pool",
        "pool": {
            "services": [
                {
                    "id": "<sustainable-backend-id>",
                    "weight": 1,
                    "priority": 1,
                    "preferredCarbonEmission": "Medium"[VV2.1][TK2.2]
                }
                {
                    "id": "<regular-backend-id>",
                    "weight": 1,
                    "priority": 1
                }
                {
                    "id": "<fallback-backend-id>",
                    "weight": 1,
                    "priority": 2
                }
            ]
        }
    }
}

If the threshold is exceeded, the gateway excludes that backend from traffic distribution, routing only to other backends with the same priority. However, if no alternatives are available, traffic will still be sent to the excluded backends to maintain continuity.

Building carbon-intelligent policies

In addition to optimising the efficiency of your backends, we are now providing visibility into the current carbon intensity of the gateway responsible for routing your traffic.

By using the new context.Deployment.SustainabilityInfo.CurrentCarbonIntensity context property, you can now build carbon-intelligent policies and adapt to your needs. This allows you to understand the carbon intensity which we have divided into multiple categories based on greenhouse gas emission intensity (gCO 2 e/kWh).

Do you want to cache data longer, reduce your telemetry volume and footprint or rate limit more aggressively? We’ve got this.

Here is an example of how you can reduce the severity of non-mission critical telemetry when the intensity is high:

<policies>
    [...]
    <inbound>
        <base />
        <choose>
            <when condition="@(context.Deployment.SustainabilityInfo.CurrentCarbonIntensity >= CarbonIntensityCategory.High)">
                <trace source="Orders API" severity="verbose">
                    <message>Lead Created</message>
                </trace>
            </when>
            <otherwise>
                <trace source="Orders API" severity="information">
                    <message>Lead Created</message>
                </trace>
            </otherwise>
        </choose>
    </inbound>
    [...]    
</policies>

Sustainable AI

As AI usage continues to rise, so does the demand for computing resources. Microsoft has responded by implementing initiatives specifically aimed at promoting sustainable AI development.

The newly introduced environmental sustainability features in Azure API Management are a significant step forward in this direction. These features are purpose-built to mitigate the environmental impact of AI by offering practical solutions that lower the carbon footprint of AI infrastructures, making them an essential tool for organisations committed to sustainable AI.

The GenAI release channel now includes these advanced sustainability capabilities, further empowering users to adopt and scale AI solutions responsibly. Interested users can access these functions after signing up for the preview.

We’d love to learn more about how we can help accelerate sustainability for your AI workloads.

Get started today

These capabilities are first being made available to classic tiers, including the GenAI release channel, with additional tiers to be included in the near future. If you want to get started, let us know and sign up here.

We look forward to understanding how our customers utilize these new capabilities and are committed to developing additional tools to help minimize the carbon footprint of your workloads.

You can learn more about:

• Environmentally sustainable APIs in Azure API Management  in our documentation and the availability per region.
• Building sustainable workloads on Microsoft Azure
• Measuring measuring carbon emissions and optimizations in Microsoft Azure
• Microsoft for sustainability

Thanks for reading,

Tom