Cloud content delivery network (CDN) stores content locally at certain edges and works with the HTTP(s) load balancing service to deliver this content to users. Google makes this happen with its global points of presence that are geographically located allowing content to be cashed closer to users. When a user requests content from a site hosted on GCP, the request arrives at an edge location of Google's network that is typically closer to the user. Cloud CDN caches at these locations, which improves the response times for the user should they request the same content again (before the CDN timeout expires). This is the CDN logo:

It is important to remember that not all data will be stored on the CDN. Whenever a user requests some content for the first time, the cache cannot fulfill the request. The cache might attempt to look for this content in a nearby cache. If it can find it, it will serve that content to the user and also store a local copy. If it does not find the content in a nearby cache, it will then send such a request to the HTTP(s) load balancer, which forwards it to one of the backend servers.

Google's caching mechanism also determines which content is more popular and ensures that it remains available while evicting content that is unpopular to conserve space. When some content is available in the cache, it is retrieved using a cache key and the cache responds directly to the user thereby giving a much faster response to the user's query.

Let's go over a high-level overview of what the cloud CDN feature looks like in GCP portal.

Click on Cloud CDN listed under the Network services. Click on Add origin on the introduction page:

An origin is your server that responds to user requests. These responses are cached by the cloud CDN. Any server from where your data originates is an origin server.