How caching works

This article provides an overview of general caching concepts and how Azure Content Delivery Network (CDN) uses caching to improve performance. If you’d like to learn about how to customize caching behavior on your CDN endpoint, see Control Azure CDN caching behavior with caching rules and Control Azure CDN caching behavior with query strings.

Introduction to caching

Caching is the process of storing data locally so that future requests for that data can be accessed more quickly. In the most common type of caching, web browser caching, a web browser stores copies of static data locally on a local hard drive. By using caching, the web browser can avoid making multiple round-trips to the server and instead access the same data locally, thus saving time and resources. Caching is well-suited for locally managing small, static data such as static images, CSS files, and JavaScript files.

Similarly, caching is used by a content delivery network on edge servers close to the user to avoid requests traveling back to the origin and reducing end-user latency. Unlike a web browser cache, which is used only for a single user, the CDN has a shared cache. In a CDN shared cache, a file request by a user can be used by another user, which greatly decreases the number of requests to the origin server.

Dynamic resources that change frequently or are unique to an individual user can't be cached. Those types of resources, however, can take advantage of dynamic site acceleration (DSA) optimization on the Azure Content Delivery Network for performance improvements.

Caching can occur at multiple levels between the origin server and the end user:

  • Web server: Uses a shared cache (for multiple users).
  • Content delivery network: Uses a shared cache (for multiple users).
  • Internet service provider (ISP): Uses a shared cache (for multiple users).
  • Web browser: Uses a private cache (for one user).

Each cache typically manages its own resource freshness and performs validation when a file is stale. This behavior is defined in the HTTP caching specification, RFC 7234.

Resource freshness

Since a cached resource can potentially be out-of-date, or stale (as compared to the corresponding resource on the origin server), it's important for any caching mechanism to control when content gets a refresh. To save time and bandwidth consumption, a cached resource isn't compared to the version on the origin server every time it's accessed. Instead, as long as a cached resource is considered to be fresh, it's assumed to be the most current version and is sent directly to the client. A cached resource is considered to be fresh when its age is less than the age or period defined by a cache setting. For example, when a browser reloads a web page, it verifies that each cached resource on your hard drive is fresh and loads it. If the resource isn't fresh (stale), an up-to-date copy is loaded from the server.

Validation

If a resource is considered stale, the origin server gets asked to validate it to determine whether the data in the cache still matches what’s on the origin server. If the file has been modified on the origin server, the cache updates its version of the resource. Otherwise, if the resource is fresh, the data is delivered directly from the cache without validating it first.

CDN caching

Caching is integral to the way a CDN operates to speed up delivery and reduce origin load for static assets such as images, fonts, and videos. In CDN caching, static resources are selectively stored on strategically placed servers that are more local to a user and offers the following advantages:

  • Because most web traffic is static (for example, images, fonts, and videos), CDN caching reduces network latency by moving content closer to the user, thus reducing the distance that data travels.

  • By offloading work to a CDN, caching can reduce network traffic and the load on the origin server. Doing so reduces cost and resource requirements for the application, even when there are large numbers of users.

Similar to how caching is implemented in a web browser, you can control how caching is performed in a CDN by sending cache-directive headers. Cache-directive headers are HTTP headers, which are typically added by the origin server. Although most of these headers were originally designed to address caching in client browsers, they're now also used by all intermediate caches, such as CDNs.

Two headers can be used to define cache freshness: Cache-Control and Expires. Cache-Control is more current and takes precedence over Expires, if both exist. There are also two types of headers used for validation (called validators): ETag and Last-Modified. ETag is more current and takes precedence over Last-Modified, if both are defined.

Cache-directive headers

Important

By default, an Azure CDN endpoint that is optimized for DSA ignores cache-directive headers and bypasses caching. For Azure CDN Standard from Edgio profiles, you can adjust how an Azure CDN endpoint treats these headers by using CDN caching rules to enable caching. For Azure CDN Premium from Edgio profiles only, you use the rules engine to enable caching.

Azure CDN supports the following HTTP cache-directive headers, which define cache duration and cache sharing.

Cache-Control:

  • Introduced in HTTP 1.1 to give web publishers more control over their content and to address the limitations of the Expires header.
  • Overrides the Expires header, if both it and Cache-Control are defined.
  • When used in an HTTP request from the client to the CDN POP, Cache-Control gets ignored by all Azure CDN profiles, by default.
  • When used in an HTTP response from the origin server to the CDN POP:

Expires:

  • Legacy header introduced in HTTP 1.0; supported for backwards compatibility.
  • Uses a date-based expiration time with second precision.
  • Similar to Cache-Control: max-age.
  • Used when Cache-Control doesn't exist.

Pragma:

  • Not honored by Azure CDN, by default.
  • Legacy header introduced in HTTP 1.0; supported for backwards compatibility.
  • Used as a client request header with the following directive: no-cache. This directive instructs the server to deliver a fresh version of the resource.
  • Pragma: no-cache is equivalent to Cache-Control: no-cache.

Validators

When the cache is stale, HTTP cache validators are used to compare the cached version of a file with the version on the origin server. Azure CDN Standard/Premium from Edgio supports both ETag and Last-Modified validators by default, while Azure CDN Standard from Microsoft supports only Last-Modified.

ETag:

  • Azure CDN Standard/Premium from Edgio supports ETag by default, while Azure CDN Standard from Microsoft doesn't.
  • ETag defines a string that is unique for every file and version of a file. For example, ETag: "17f0ddd99ed5bbe4edffdd6496d7131f".
  • Introduced in HTTP 1.1 and is more current than Last-Modified. Useful when the last modified date is difficult to determine.
  • Supports both strong validation and weak validation; however, Azure CDN supports only strong validation. For strong validation, the two resource representations must be byte-for-byte identical.
  • A cache validates a file that uses ETag by sending an If-None-Match header with one or more ETag validators in the request. For example, If-None-Match: "17f0ddd99ed5bbe4edffdd6496d7131f". If the server’s version matches an ETag validator on the list, it sends status code 304 (Not Modified) in its response. If the version is different, the server responds with status code 200 (OK) and the updated resource.

Last-Modified:

  • For Azure CDN Standard/Premium from Edgio only, Last-Modified is used if ETag isn't part of the HTTP response.
  • Specifies the date and time that the origin server has determined the resource was last modified. For example, Last-Modified: Thu, 19 Oct 2017 09:28:00 GMT.
  • A cache validates a file using Last-Modified by sending an If-Modified-Since header with a date and time in the request. The origin server compares that date with the Last-Modified header of the latest resource. If the resource hasn't been modified since the specified time, the server returns status code 304 (Not Modified) in its response. If the resource has been modified, the server returns status code 200 (OK) and the updated resource.

Determining which files can be cached

Not all resources can be cached. The following table shows what resources can be cached, based on the type of HTTP response. Resources delivered with HTTP responses that don't meet all of these conditions can't be cached. For Azure CDN Premium from Edgio only, you can use the rules engine to customize some of these conditions.

Azure CDN from Microsoft Azure CDN from Edgio
HTTP status codes 200, 203, 206, 300, 301, 410, 416 200
HTTP methods GET, HEAD GET
File size limits 300 GB 300 GB

For Azure CDN Standard from Microsoft caching to work on a resource, the origin server must support any HEAD and GET HTTP requests and the content-length values must be the same for any HEAD and GET HTTP responses for the asset. For a HEAD request, the origin server must support the HEAD request, and must respond with the same headers as if it received a GET request.

Default caching behavior

The following table describes the default caching behavior for the Azure CDN products and their optimizations.

Microsoft: General web delivery Edgio: General web delivery Edgio: DSA
Honor origin Yes Yes No
CDN cache duration Two days Seven days None

Honor origin: Specifies whether to honor the supported cache-directive headers if they exist in the HTTP response from the origin server.

CDN cache duration: Specifies the amount of time for which a resource is cached on the Azure CDN. However, if Honor origin is Yes and the HTTP response from the origin server includes the cache-directive header Expires or Cache-Control: max-age, Azure CDN uses the duration value specified by the header instead.

Note

Azure CDN makes no guarantees about minimum amount of time that the object will be stored in the cache. Cached contents may be evicted from the CDN cache before they are expired if the contents are not requested as frequently to make room for more frequently requested contents.

Next steps