In the ongoing battle to attract and retain customers online, more businesses are turning to site personalisation to deliver dynamic, relevant content that can be tailored to enrich user experience. The process, however, can quickly bring servers to their knees.
In a multitiered Web architecture, frequent calls to application servers and databases place enormous strain on back-end resources and cause throughput bottlenecks, inducing companies to invest in additional hardware and bandwidth to keep pace with demand. Dynamic content acceleration software, which works by caching reusable data, is finally offering relief to businesses in need of speedier personalised content delivery.
How can a Web page be both reusable and tailored to an individual user? Dynamic content acceleration relies on the expectation that even personalised content will call for data that can be recycled. A content accelerator sits between your Web server and your back-end resources to intercept incoming script requests, such as those from Java and Active Server Pages, CGI, and servlets. It completes the HTML response page using cached data such as frequently requested catalog products, articles, news items, and stock prices.
Similar to the way static-page caching can solve network latency problems, dynamic content acceleration reduces server-side latency by decreasing the number of high-overhead calls to application servers and databases needed to fill a request. The software-based approach improves e-commerce transactions without the need to scale up costly back-end resources.
The number of companies offering well-designed solutions in this nascent field is slim. Among the forerunners, albeit through different implementation strategies, are SpiderCache, Persistence Software with Dynamai, and Oracle with its Edge Side Includes (ESI) initiative.
One of the most compelling dynamic content acceleration solutions can be found in PreLoader 3 from Chutney Technologies, which we recently tested.
Chutney takes dynamic caching to the next level by caching individual logic on the page rather than by caching an entire HTML page response, which most other products do.
By breaking pages into discreet components, PreLoader stores the results of partially constructed page elements and draws from these pieces to create new page combinations. For example, when a client requests a server-side script from the Web server, PreLoader intercepts and processes the call. PreLoader recognises the tagged code, checks to see whether or not it has comparable results in its cache, and uses these items to construct the final HTML page. The process is completed without reinvolving the application server and reduces I/O calls to the database, thereby reducing the possibility of throughput bottlenecks.
PreLoader's component-level caching offers speed advantages over page-level caching, thanks to the finer granularity of its reusable HTML pieces. Although Oracle's nonscripted ESI solutions may be far easier to implement, PreLoader boasts the additional benefit of working in any Web or application server environment.
In such dynamic implementations, a top priority is ensuring the caching process doesn't undermine the relevancy of the information. Outdated pages can yield unexpected results, demanding the cache be updated regularly with current data. PreLoader uses several methods of optimization, including Chutney's own predictive probability algorithm to ensure maximised cache effectiveness.
PreLoader allows multiple caches to be used to create and store different data types, optimising access efficiency. An additional feature for granting prioritized access to specific users or groups such as A-list customers can even help provide targeted performance benefits.
Because the number of implementations is still relatively limited and the products themselves remain relatively untested, dynamic content acceleration may find all but the most demanding sites waiting for further development. Within the next year, however, most static cache vendors likely will begin to incorporate some form of dynamic cache capability into their upgrades.
With the risk factor in implementation so minimal, however, we recommend that any company with a bottom line riding on personalized content delivery take out some insurance by investing in dynamic content acceleration.
Chutney Technologies PreLoader 3.0
Chutney PreLoader makes it possible to break apart scripts, reduce their logic to taggable components, and associate output to the actual HTML elements on the page.
PreLoader runs on Windows NT, Windows 2000, Linux, and Solaris, and it offers a Java-based graphical administrative interface for attaching to and monitoring dynamic cache instances. We found the feature useful for optimizing performance results.
Furthermore, the application is available in several enterprise-worthy configurations that sport features such as fail-over and clustering capabilities for high-availability scenarios.
Although using PreLoader's pretagged scripts leads to favorable speed and cache hit ratios, the process demands significant reconfiguration of your site's source code. Not only is the process error-prone for already-operational site logic, but it brings additional development expense to a price tag already two to three times that of its competitors.
The Active Server Pages version we tested did not include Chutney's Automated Script Analyzer and Processor tool. The tool reportedly helps reduce the complexities of the tagging process and aids in optimizing cache usage.
We strongly recommend that you consider PreLoader for enterprise-strength deployment. All told, we found Chutney's component-based approach a good match for reducing the cost of back-end technology investments in personalised, highly trafficked sites. But you may not want to tackle the tagging on your own.