Starting today, you can deploy Amazon CloudFront, our global content delivery network (CDN), in front of your gRPC API endpoints.

gRPC is a modern, efficient, and language-agnostic framework for building APIs. It uses Protocol Buffers (protobuf) as its interface definition language (IDL), which enable you to define services and message types in a platform-independent manner. With gRPC, communication between services is achieved through lightweight and high-performance remote procedure calls (RPCs) over HTTP/2. This promotes efficient and low-latency communication across services, making it ideal for microservices architectures.

gRPC offers features such as bidirectional streaming, flow control, and automatic code generation for a variety of programming languages. It’s well-suited for scenarios in which you require high performance, efficient communication, and real-time data streaming. If your application needs to handle a large amount of data or requires low-latency communication between client and server, gRPC can be a good choice. However, gRPC might be more challenging to learn compared to REST. For example, gRPC relies on the protobuf serialization format, which requires developers to define their data structures and service methods in .proto files.

I see two benefits of deploying CloudFront in front of your gRPC API endpoints.

First, it allows the reduction of latency between the client application and your API implementation. CloudFront offers a global network of over 600+ edge locations with intelligent routing to the closest edge. Edge locations provide TLS termination and optional caching for your static content. CloudFront transfers client application requests to your gRPC origin through the fully managed, low-latency, and high-bandwidth private AWS network.

Secondly, your applications benefit from additional security services deployed on edge locations, such as traffic encryption, the validation of the HTTP headers through AWS Web Application Firewall, and AWS Shield Standard protection against distributed denial of service (DDoS) attacks.

Let’s see it in action
To start this demo, I use the gRPC route-guide demo from the official gRPC code repository. I deploy this example application in a container for ease of deployment (but any other deployment option is supported too).

I use this Dockerfile

FROM python:3.7
RUN pip install protobuf grpcio
COPY ./grpc/examples/python/route_guide .
CMD python route_guide_server.py
EXPOSE 50051

I also use the AWS Copilot command line to deploy my container on Amazon Elastic Container Service (Amazon ECS). The Copilot command prompts me to collect the information it requires to build and deploy the container. Then, it creates the ECS cluster, the ECS service, and the ECS task automatically. It also creates a TLS certificate and the load balancer for me. I test the client application by modifying line 122 to use the DNS name of the load balancer listener endpoint. I also change the client application code to use grpc.secure_channel instead of grpc.insecure_channel because the load balancer provides the application with an HTTPS endpoint.

When I’m confident my API is correctly deployed and working, I proceed and configure CloudFront.

First, in the CloudFront section of the AWS Management Console, I select Create distribution.

Under Origin, I enter my gRPC endpoint DNS name as Origin domain. I enable HTTPS only as Protocol and leave the HTTPS port as is (443). Then I choose a Name for the distribution.

Under Viewer, I select HTTPS only as Viewer protocol policy. Then, I select GET, HEAD, OPTIONS, PUT, POST, PATCH, DELETE as Allowed HTTP methods. I select Enable for Allow gRPC requests over HTTP/2.

Under Cache key and origin requests, I select AllViewer as Origin request policy.

The default cache policy is CacheOptimized, but gRPC isn’t cacheable API traffic. Therefore, I select CachingDisabled as Cache policy.

AWS WAF helps protect you against common web exploits and bots that can affect availability, compromise security, or consume excessive resources. For gRPC traffic, AWS WAF can inspect the HTTP headers of the request and enforce access control. It doesn’t inspect the request body in protobuf format.

For this demo, I choose to not use AWS WAF. Under Web Application Firewall (WAF), I select Do not enable security protections.

I also keep all the other options with their default value. HTTP/2 support is selected by default. Do not disable it because it is required for gRPC.

Finally, I select Create distribution.

There is only one switch to enable gRPC on top of the usual setup. When turned on, with HTTP/2 and HTTP POST enabled, CloudFront detects gRPC client traffic and forwards it to your gRPC origin.

After a few minutes, the distribution is ready. I copy and paste the endpoint URL of the CloudFront distribution, and I change the client-side app to make it point to CloudFront instead of the previously created load balancer.

I test the application again, and it works.

Pricing and Availability
gRPC origins are available on all the more than 600 CloudFront edge locations at no additional cost. The usual requests and data transfer fees apply.

Go and point your CloudFront origin to a gRPC endpoint today.

— seb