Ensuring Fresh Content

Ensuring Fresh Content

A requirement for any caching system is the ability to ensure that users see the same content from a network cache as they would from the Web. Every Web page comprises several Web objects and each Web object has its own caching parameters, determined by content authors and HTTP standards (see the “HTTP Caching Standards” section). Thus, even a Web page with real-time objects typically has many other objects that are cacheable. Rotating ad banners and Common Gateway Interface (CGI)-generated responses are examples of objects that are typically noncacheable. Toolbars, navigation bars, GIFs, and JPEGs are examples of objects that are typically cacheable. Thus, for a given Web page, only a few dynamic objects need to be retrieved from the end server, while static objects can be fulfilled locally.

Cisco Cache Engine products deliver fresh content by obeying the HTTP caching standards and by enabling cache administrators to have control over when content should be refreshed from origin Web servers.

HTTP Caching Standards

HTTP 1.0 and 1.1 are caching standards, which specify caching parameters for each object on a Web page.

HTTP 1.0 allows content authors to enable a “Pragma: no cache” header field for any object that should not be cached and allows authors to enable content to be cached indefinitely.

HTTP 1.1 allows content authors to specify how long content is to be cached. For each object on a Web page, content authors can choose among the following caching attributes:

• Noncacheable

• OK to cache (the default setting)

• Explicit expiration date

HTTP 1.1 has a freshness revalidation mechanism called If-Modified-Since (IMS) to ensure that cached data is up to date. A cache engine will send a lightweight IMS request to the end Web server when the cache engine receives requests for cached content that has expired or IMS requests from clients where the cached content is more than a configured percentage of its maximum age. If the object has not been modified on the end server since the object was cached, the end server will return a lightweight message indicating that the cache engine can deliver its cached copy to clients. If the object has been modified on the end server since the object was cached, the end server will return this information to the cache engine. If the case of the client issuing an IMS request, and the content is less than a configured percentage of its maximum age, the cache will serve the content without checking if it is fresh.

Cache Engine Content Freshness Controls

Administrators can control the freshness of Web objects in a cache engine by configuring a parameter called the freshness factor, which determines how fast or slow content expires. When an object is stored in the cache, its time-to-live (TTL) value is calculated using the following formula:

TTL value = (Current date - last modified date) * Configurable freshness factor

When an object expires, based on its TTL value, the cache engine will issue an IMS request the next time the object is requested (see “HTTP Caching Standards” section for a description of the IMS process). If an administrator wants to adopt a conservative freshness policy, he or she can set the freshness factor to a small value (such as 0.05), so that objects expire more quickly. But the disadvantage to this approach is that IMS requests will be issued more frequently, consuming extra bandwidth. If an administrator wants to adopt a liberal freshness policy, the fresh factor can be set to a larger value, so that objects will expire more slowly and the IMS bandwidth overhead will be smaller.

Browser Freshness Controls

Finally, clients can always explicitly refresh content at any time by using the browser's reload/refresh button.

The reload/refresh command is a browser-triggered command to request a data refresh. A reload/refresh will issue a series of IMS requests asking for only data that has changed.

The shift+reload/shift+refresh command is an extension of the reload/refresh command. In correctly implemented browsers, this command always triggers a “pragma: no cache” rather than an IMS request. As a result, cache engines are bypassed and the end server directly fulfills all content.

Summary

Much of the traffic on the Web is redundant, meaning that users in the same location often access the same content over and over. Eliminating a significant portion of recurring telecommunications offers huge savings to enterprise and service providers.

Caching is the technique of keeping frequently accessed information in a location close to the requester. The two key benefits are:

• Cost

• Improved usability

Implementing caching technology in a network accelerates content delivery, optimizesWAN bandwidth, and enables content monitoring.

Cisco has created a network-integrated cache engine by pairing system-level software and hardware.

Review Questions

Q—On what concept is network caching based?

A—Based on the assumption that users access the same content over and over.

Q—What are two secondary benefits of implementing caching technology?

A—

1. Secure access and control.

2. Operational logging—administrators can log how many hits sites receive.

Q—Provide a brief description of network-integrated caching technology.

A—Network-integrated caching technology combines system-level software and hardware. Network-integrated caches must be managed like network equipment, designed like high-density hardware, and transparently inserted into the network.

Q—How do Cisco cache engines ensure that web pages are kept up to date?

A—By obeying HTTP caching standards that dictate which elements on a page can be cached and which cannot. Those that are not are retrieved from the source every time they are accessed.

Q—Name an object that can be saved in cache memory, and one that cannot.

A—Saved in cache: rotating banners, GIFs and JPEGs, toolbars, navigation bars. Noncacheable: CGI-generated responses.