On this page

Run Anywhere - Deploy your Gateway

Once you configured, and tested your gateway. Now, it is time to deploy it. Hive Gateway Runtime uses Web Standards (WHATWG Fetch API) not only as an HTTP client but also for handling the server-side. That gives us the ability to run the gateway in any environment that runs JavaScript.

Node.js is the most common server-side environment in JavaScript ecosystem but it doesn’t use Web Standards for handling HTTP requests. So we use a library called @whatwg-node/server that allows us to create a wrapper between node:http and Fetch API.

Programmatic Usage

In all of the deployment examples listed, except for Docker and Google Cloud Run, you will be using the Hive Gateway programmatically through its programmatic API.

The runtime of Hive Gateway, our recommended way of using the gateway programmatically comes in a separate package @graphql-hive/gateway-runtime.

npm i @graphql-hive/gateway-runtime

Improving the experience of Hive Gateway in various JavaScript environments (especially serverless/on-the-edge), the @graphql-hive/gateway-runtime comes in a slimmer version set up for optimized runtime and smaller size. This leads to a difference in the configuration options compared to the CLI version @graphql-hive/gateway.

See the Hive Gateway Configuration Rereference for more details about these differences.

Other Environments (Not Listed)

Let’s say you have an environment that is not listed here, you can still deploy your Hive Gateway using its programmatic API. In this case, we will show here how to pass the request information from your environment to Hive Gateway, then get the response for your environment back.

import { createGatewayRuntime } from "@graphql-hive/gateway-runtime";
import type {
  ImaginaryEnvironmentRequest,
  ImaginaryEnvironmentServerContext,
} from "@imaginary-environment/types";
import { getMySupergraph } from "./my-supergraph.js";

// Initialise and configure the Hive Gateway runtime
const gateway = createGatewayRuntime<ImaginaryEnvironmentServerContext>({
  supergraph: () => getMySupergraph(),
});

// Let's say it needs a function exported
export async function gatewayEndpoint(
  envRequest: ImaginaryEnvironmentRequest,
  envContext: ImaginaryEnvironmentServerContext,
) {
  // The runtime provides a fetch function which has exactly the same signature with regular `fetch`
  const res = await gateway.fetch(
    envRequest.url,
    {
      method: envRequest.method,
      headers: envRequest.headers,
      body: envRequest.body, // Body can be a string or a ReadableStream or UInt8Array, see [BodyInit](https://developer.mozilla.org/en-US/docs/Web/API/BodyInit)
    },
    envContext,
  );

  // You can create an object from [`Headers`](https://developer.mozilla.org/en-US/docs/Web/API/Headers) object
  const headersObj: Record<string, string> = {};
  res.headers.forEach((value, key) => {
    headersObj[key] = value;
  });
  // It returns [`Response`](https://developer.mozilla.org/en-US/docs/Web/API/Response) object
  // See the methods and properties of the Response object from the link
  // You can get a string
  const bodyText = await res.text();
  // You can get a stream
  const bodyStream = res.body;
  // You can get a buffer
  const bodyBuffer = await res.arrayBuffer();
  // You can get a JSON object
  const bodyJson = await res.json();
  // You can get a blob
  const bodyBlob = await res.blob();
  // You can get a form data
  const bodyFormData = await res.formData();

  // Then you can return the response to your environment
  return {
    status: res.status,
    statusText: res.statusText,
    headers: headersObj,
    bodyText,
  };
}