Speed up your dynamic Web content

Web-based content is increasingly becoming dynamic rather than static. This means Web pages are being generated on demand with content tailored to each user.

A simple example is a personal greeting that pops up when a regular customer returns to a Web site. A more elaborate scheme might provide a customer with a set of recommendations based on past purchases or site interactions.

---

How it works
Subscribe to the Tech Update newsletter
  Here is a weekly newsletter to help you stay abreast of new networking standards and technologies by providing down-to-earth explanations of how they work.

---

While dynamic sites provide a far richer experience for users than static sites and have been shown to increase customer retention, generating Web pages on the fly exerts a major toll on server resources. The results are bottlenecks at the server and slower response times for end users.

To address the latencies inherent in dynamic sites, a new class of product is emerging called dynamic content accelerators. These server-side accelerators intelligently feed data to application servers, letting pages be created much faster than they are on nonoptimized sites.

Basics of content delivery

When a user types in a URL to request a Web page, the page is created by an application server, which executes a script that builds the page. This script contains actions such as calls to database systems to retrieve content. The result is an HTML "blueprint" for the requested page.

The page is then delivered back to the browser in a quick, text-only, nonbandwidth-intensive transfer that does not incorporate graphics. Finally, the browser must fetch the graphics, requesting each object from the appropriate server address based on the embedded URLs in the HTML page.

Because browsers are limited to downloading two to four objects at a time and a typical Web page may contain 30 to 40 embedded objects, a good deal of back-and-forth handshaking between the browser and server is required to complete the loading of a page.

During the past few years, object delivery has been successfully optimized through network caching. By storing and serving objects from the network edge, caching slashes the time it takes a browser to load an object. Caching can be deployed as a product (CacheFlow and Inktomi are leading vendors) or as a service (Akamai leads this category).

But caching addresses network latency, not server latency. Caching doesn't effectively address dynamic page generation, which typically accounts for 40% of the time required to deliver a Web page.

The problem is complex, involving a mix of business logic execution, database and/or file system access (from local or remote systems), and content transformation and formatting, such as converting XML to HTML. Each of these tasks requires considerable server resources, creating load issues. Until the page is created, subsequent steps in the process are stalled.

Emerging alternative

Dynamic content accelerators take advantage of the fact that much of the content in a dynamically generated page is reusable. A dynamic content accelerator caches individual page "components" for faster access. A component is a group of data that is displayed together on a page, such as a set of top news stories or a product's price and attributes.

For each component on a page, the application server makes a request to the dynamic content accelerator. If the data resides in the accelerator's RAM-based cache, it is instantly returned to the application server in ready-to-display HTML format, bypassing the processing and I/O tasks typically required to create the component.

Sites gain three key efficiencies by caching dynamic page content:

? Script routines do not have to be executed.

? Data elements do not have to be retrieved from local or remote database systems.

? Data elements do not have to be converted from formats such as XML or wireless markup language into HTML. Servers and storage subsystems are significantly off-loaded through this approach.

With dynamic content accelerator technology, Web pages are created and delivered to users' browsers in a fraction of the normal time, and the graphics-fetching process can begin. Site scalability is dramatically enhanced as requests for repeatedly accessed content are fulfilled through a high-performance caching engine.

---

Govatos is vice president of Chutney Technologies, an e-business software infrastructure provider. He can be reached at greg.govatos@chutneytech.com

How XML enables dynamic content
Network World, 03/13/00.