Open Link concurrency and scaling docs with resource requirements (#67)

Author: enisdenjo

packages/documentation/content/docs/gateway/concurrency-and-scaling.mdx

@@ -142,6 +142,12 @@ That said, forking does work on Bun since it implements the Node.js `cluster` mo
 descriptors cannot be passed between workers, which limits TCP server load-balancing across
 processes to Linux only.
 
+## Sizing Your Deployment
+
+Once you have chosen a fork count, head over to
+[Resources Requirements](/docs/gateway/deployment/resources-requirements) to translate that decision
+into concrete CPU and memory allocations for your pods or VMs.
+
 ## Scaling Considerations for Multiple Instances
 
 When running multiple gateway workers (via forking) or multiple gateway instances (e.g. multiple

packages/documentation/content/docs/gateway/deployment/resources-requirements.mdx

@@ -2,6 +2,15 @@
 title: Resources Requirements
 ---
 
+import { Callout } from "@hive/design-system/hive-components/callout";
+
+<Callout>
+  Before sizing your deployment, read [Concurrency &
+  Scaling](/docs/gateway/concurrency-and-scaling) first. The CPU and memory
+  numbers below depend directly on how many workers you run - understanding
+  forking and its memory implications will make this guide much more actionable.
+</Callout>
+
 Defining the right amount of resources for your GraphQL gateway is crucial for maintaining
 performance and reliability. This guide will help you understand how to allocate resources
 effectively.
@@ -31,11 +40,13 @@ Set `NODE_ENV` to `production` to ensure your gateway runs in production mode. T
 enables [parallelism optimizations](https://nodejs.org/api/cluster.html) that can enhance
 performance.
 
-By default, the gateway uses the available parallelism (number of CPUs, based on NodeJS's
-`os.availableParallelism`) to determine the number of workers to spawn.
+By default, the gateway runs as a single process with no forking. You can set the `FORK` environment
+variable to specify the number of workers to spawn. For example, to spawn 4 workers, set `FORK=4`.
+Each worker is an independent Node.js process with its own memory heap, so your total memory
+requirement scales linearly with the fork count.
 
-You can set the `FORK` environment variable to specify the number of workers you want to spawn. For
-example, to spawn 12 workers, set `FORK=12`.
+See [Concurrency & Scaling](/docs/gateway/concurrency-and-scaling) for a full explanation of how
+forking works, how to choose the right fork count, and how to plan memory accordingly.
 
 Consider unsetting the `DEBUG` env variable if you don't need debug logs. This can have significant
 impact on CPU usage.