5.1 Site requirements
Requirements are key to any software project and it is no different with creating a WebSphere Commerce site that can benefit from the caching capabilities of DynaCache. These requirements can help you drive the design of your WebSphere Commerce implementation so that its components can be cached at the correct level of granularity, and can also help to create an effective cache policy.
Listed here are some of the key performance requirements:
| Concurrent users: The total number of active users that are expected to be using your e-commerce site. A user may visit your site many times but can have only one session at a time. When considering a requirement for concurrent users, an important metric is think-time, which refers the time between requests for any given user. As you increase think-time, the number of supported concurrent users will increase and think-time will vary by browse scenario. |
|
| Peak versus average usage: The variance in the load that is being placed on your e-commerce site. This can depend on a variety of conditions, for example, time of day, time of the year, or special events. Performance requirements have to be maintained during peak usage as well as during average usage. |
|
| Page views per second: Throughput in terms of the total number of times a user visits or views a page, measured per second and scaled over multiple users. A page view includes requests for all the files that are contained on a page as well as the page itself. A page view can include one or more hits, where a hit is any type of request to the server. |
|
| Response time: The amount of time required to complete a single page view or page hit, and measured in seconds. Response time will depend heavily on other requirements and caching needs to be taken into account to optimize response time. |
|
| Kilobytes per second: Throughput in terms of the total number of bytes that can be transferred to a user per unit of time, usually seconds. Caching can help to meet this type of requirement, especially when bandwidth is limited. |
|
| Browse-buy ratio: The ratio of the number of users who are visiting your e-commerce site but not buying products versus the number of users who are buying products, meaning checking out from their shopping cart. Typically, a B2B e-commerce site will have a smaller browse-buy ratio than a B2C site. For example, a B2C browse-buy ratio might be 100:2 or 2 percent buyers while a B2B site might have a browse-buy ration of 100:35 or 35 percent buyers. In scenarios where the browse-buy ratio is extremely high, for example in B2C sites, caching can play an especially crucial role in improving performance since it is more likely that we can cache pages related to browsing versus pages related to buying. In WebSphere Commerce, we cache the catalog pages that many users will browse through rather than the checkout pages, which are unique to a single users at a single instance. |
|
| Locale: The location from which users will be browsing your Web site. This is particularly significant if your Web site must support a variety of locales and the amount of traffic from each locale may be known. For example, if your site must support both English and French, but you know that 95% of your visitors are English speaking, then clearly caching the English pages will yield a greater benefit than caching the French ones. |
When creating an e-commerce site, there are a variety of performance requirements that the site must meet. Each of these requirements demand that performance be fine tuned, including a strategy for caching to improve performance. Furthermore, capacity planning models created and used by IBM assume that any WebSphere Commerce implementation will implement caching using DynaCache.
E-commerce applications must always be concerned with performance, especially in the case of B2C WebSphere applications, and as a result the use of DynaCache in these scenarios is not optional.