Cleanly separating CDN caching and browser caching

Which cache rules differ between the CDN and the user's browser?

Browser caches are user-specific, while CDNs share responses between requests. Therefore, cache control, S-maxage, private, Vary, and authentication must be consciously combined for each piece of content.

Variant scope

1. Organize response classes as public, private, versioned, and dynamic.

2. Define browser and shared cache rules, as well as relevant variants, separately.

3. Test hits, revalidation, and private responses in the browser, CDN, and Origin.

Storage instance

• Personalization Private or logged-in content is not stored in a shared cache.

Data mixing

• Data mixing A CDN delivers a personalized response to another request, thereby exposing session- or user-related content.

• Stale browsers The CDN is cleared, but long browser runtimes retain old HTML content.

• Vary explosion Unlimited header values ​​destroy hit rates and predictability because almost every request receives its own cache key.

Use case: "Data mixing"

A versioned script may remain in the browser and CDN for an extended period, while the HTML page revalidates briefly. A logged-in account area has the directive `Cache-Control: private`; a CDN hit for this content is considered an error, even if the response is technically fast.

Personalization

• Cache hits, misses, and bypasses at each browser, CDN, and origin level, as well as by response class.

• Incorrectly shared or unexpectedly old responses based on header combinations.

What questions remain after "Controlling CDN and Browser Caching Separately"

Maintaining Correct File Permissions and Ownership on Web Servers delves deeper into the "Storage Instance" checklist. The key question is: What ownership and permissions distribution protects web server files without blocking operation?

A complementary perspective is offered Define cache runtimes based on file type and change risk.It answers the question: "How do you define cache runtimes based on file type and change risk?"

If you want to practically implement "Controlling CDN and Browser Caching Separately," you can refer to Robust Website Systems This document focuses on "HTTP and CDN Caching" and "Storage Instance."

Conclusion: Control CDN and Browser Caching Separately

Browsers and CDNs have different trust and life spaces. Good headers explicitly define these differences and prevent the sharing of private responses.

Sources and Further Information

The classification of "controlling CDN and browser caching separately" is based on the following official documentation and standards.

• RFC 9110: HTTP SemanticsThe standard provides the semantics for methods, status codes, fields, validators, variants, and intermediaries.

• RFC 9111: HTTP CachingThe Internet standard defines storage, freshness, revalidation, directives, and security requirements for HTTP caches.