Every time we make a phone call and get a busy signal, we assume the person we’re calling is on the phone – we don’t even think about the possibility that the telephone network isn’t working. Not so with the Net. The Internet is carrying a growing amount of traffic and carrying it greater distances. Too bad it’s not as reliable as the telephone network. Fortunately, there are technologies that can help both traditional and personalized web sites overcome some of the congestion on the Internet.
You have probably heard people say, “I can’t get to that site – their server must be down.” The fact is that the server is probably not down, but parts of the Internet are so congested that browsers can’t connect to the server.
As traffic on the Internet has grown, network managers have increased the number of connection points in an attempt to route traffic more efficiently. Unfortunately, when a connection is overloaded, it can keep some users from being able to reach a web site, while other users going through different connections may not experience a problem.
In general, the way to reduce congestion is to reduce the distance between the user and the server. The cell phone network knows which cell should handle a call, and a similar technique can be used on the Internet. With web users located throughout the world, a set of servers throughout the world needs to be smart enough to know which server should respond to the user.
This technique, called caching, is typically used to serve content that doesn’t change, such as static web pages, graphics and downloadable files.
One approach to caching is to serve content from servers geographically close to web users. The caching servers don’t need to request updates from the central server because static files don’t change often.
This reduces the number of Internet connections between the server and the user, which speeds content delivery and reduces the load on the Internet. The main benefit of caching is that files are delivered to the viewer more quickly and with fewer delays. In addition, the main server is taxed less, which can reduce the cost and complexity of that datacenter.
For example, a user in San Francisco who accesses a web page or graphic from a web site hosted in Boston without caching would retrieve all of the files from the server in Boston. When the same user in San Francisco accesses a Boston web site that is cached in Los Angeles, the files travel a much shorter distance through fewer connection points.
Web-hosting companies with large networks and multiple datacenters, such as GTE Internetworking, offer caching servers at various datacenters in their network. Their caching servers automatically update themselves when files are updated, which takes the load off the central server or cluster of servers.
Even though caching servers are used to speed delivery of static files, some personalized web sites can use caching, too. Sites that dynamically create links to static pages can use caching to serve the non-changing files on the site.
Another use for caching is to improve the performance of a cluster of servers such as those used at high-traffic commerce and personalized sites. Sites that serve dynamically changing content have not been candidates for caching. Now, technology is available from companies such as TimesTen that can cache changing data, such as shopping cart information or personalized web pages. Instead of caching files at servers around the world, TimesTen’s technology stores pieces of content at the main datacenter to help a cluster of servers retrieve dynamic content more quickly.
Caching is not always easy to implement, but it is worth considering for a high-traffic site. With proper planning and technical implementation, cache management can overcome some of the problems of lost traffic, customers, and sales.