Introduction to cache policies - Edge Security Acceleration

Edge Security Acceleration (ESA) improves your website performance by caching your static resources on a distributed network of points of presence (POPs). This reduces requests to your origin server, lowers latency, and lets you dynamically adjust caching policies to fit your business needs.

Caching benefits

ESA boosts your cache hit rate with policies like ignoring or sorting query string parameters. This ensures that requests for identical content are served from the cache, reducing redundant origin requests.
You can dynamically adjust cache policies for different scenarios, such as bypassing the cache during debugging or canary releases.
Fewer requests to your origin server lower latency, improve the user experience, and reduce the load on your origin server, which increases its stability and performance.

Default rules and development mode

Default cache rules automatically cache static resources, such as images and CSS/JS files, on POPs. This reduces pressure on your origin server and improves access speed. Development mode lets you temporarily bypass cache components to access the origin server directly. This is useful for debugging or canary releases.

Enable development mode: Bypass cache components to access the origin server directly. This is useful for canary releases or debugging.
Note
Enabling this feature can significantly increase back-to-origin traffic. Disable development mode after you finish debugging.
Default cache rules: ESA automatically caches static resources, such as images and CSS/JS files, on edge POPs to prioritize user requests.

Set the cache TTL

Configure the browser and edge cache time-to-live (TTL) to improve access speed and conserve server resources. This leads to more efficient network management and optimization.

Browser cache TTL: By default, the TTL follows the origin server's Cache-Control header. You can override this setting. For example, set the TTL for static resources to 30 days. Configure a shorter TTL for frequently updated resources. For cold data (infrequently accessed data) that does not change often, configure a longer TTL.
Edge cache TTL: Configure the TTL by directory (such as /images/*) or file extension (such as .jpg,.css). For frequently updated resources, set the TTL to 3600 seconds. For infrequently accessed resources, set the TTL to 86400 seconds.

Increase the cache hit ratio

You can increase the cache hit ratio by ignoring or sorting query strings. Ignoring query strings treats requests with different query parameters as the same resource for caching. Sorting query strings treats requests with parameters in a different order as the same resource after sorting.

Ignore query strings: Enable the query string feature to cache requests that have the same URL but different ?param=value parts as a single resource. For example, example.com/index.html?v=1.0 and example.com/index.html?v=2.0 are treated as the same resource.
Sort query strings: Enable the query string sorting feature. The system then treats requests with the same parameters in a different order as the same cache key. For example, /api?a=1&b=2 and /api?b=2&a=1 are treated as the same cache key.

Advanced features

ESA provides tiered cache and cache reserve to help you optimize resource use, improve user experience, lower costs, and reduce bandwidth pressure on your origin server.

Tiered cache: Tiered cache is an advanced performance optimization technique. ESA caches origin resources on tiered cache nodes worldwide. This allows data to be served from the POP closest to the user, which improves overall system performance and efficiency.

Cache reserve: When enabled, ESA directs origin requests for cold traffic (infrequently accessed requests) to a specified cache reserve instance. This prevents the resources from being frequently evicted due to low popularity. This feature is suitable for infrequently accessed resources that require long-term storage. Cache reserve POPs reserve dedicated disk space to ensure files are cached for the specified time. This avoids unnecessary egress traffic fees from the origin server and reduces bandwidth pressure on the origin server.