InformationWeek
Defining The Business Value Of Technology
|
September 25, 2000 |
 |
Web-Site Scalability
Business Drives Web Scalability
The goal of round-the-clock service for online businesses isn't impossible, though it can be hard to reach. Scalability--a Web site's ability to give users stable response times, handle transaction integrity, and be fault-tolerant--must be planned from the start.
 |
| More on scalability: |
|
|
| TechEncyclopedia: |
 Send Us Your Feedback |
raditional companies moving toward the Web and 100% online businesses all strive for more scalable sites. Both want to offer visitors around-the-clock service. This goal isn't entirely unattainable, but reaching it will be a major challenge. Many different parameters must be taken into consideration.These parameters can be grouped into two categories: technical components--such as hardware, application software, and networks--and Web-site traffic. The latter is always welcome but hardly controllable, so let's concentrate on the former.
The ability to scale incrementally to meet evolving customer demand, as well as the ability to meet spikes in user demand quickly, are equally important. But what exactly is scalability? It's a combination of many factors: a Web site's ability to provide users with stable response times, handle transaction integrity, and be fault-tolerant when an outage occurs.
Scalability is all about reliability and performance. IT professionals seem to agree that scalability is, above all, a design issue. Scalability issues are encountered on all design levels. The first level consists of a network hookup to an E-business system's intranet. A Web server and an application server, including a database-management system or back-end systems such as legacy or enterprise resource planning systems, represent the second level. Finally, the third level is made up of a local network's hardware servers and network equipment. The overall design of a scalable system must take all these aspects into consideration.
In theory, distributed computing and redundancy are the two concepts behind scalable solutions. These concepts don't exclude one another; on the contrary, the best scalable systems use both.
Redundancy means having two elements that provide the same feature within the same system, one of which is designated as a backup to the first. Ideally, a global system will use both elements when everything is working normally, then switch to one should the other crash. Redundancy was popularized by servers from Tandem (now a part of Compaq) on the server side, and by distributed tools from Forté (now owned by Sun Microsystems) during the three-tier client-server era, which marked the beginning of distributed computing. Today, fault-tolerant hardware is part of most manufacturers' offerings.
Distributed computing is represented by three standards: the Object Management Group's Common Object Request Broker Architecture (Corba), Microsoft's Distributed Component Object Model and Digital Network Architecture, and Sun's Enterprise JavaBeans, which is part of the Java 2 Enterprise Edition platform. The Internet economy has dramatically increased the need for scalable systems; until now, not even the largest companies had to handle such large numbers of users.
This demand has also sped up the evolution of distributed computing. Considered a luxury architecture not so long ago, the Internet has made the distributed computing model standard. Indeed, two technical implementations of distributed computing dominate Web applications: Microsoft DCOM-DNA and Sun's Enterprise JavaBeans and the Java 2 Enterprise Edition.
With its Windows 2000 platform, Microsoft has emphasized component distributions via its COM+ services. As for Sun, the company has always been very successful on the server side with Java 2 Enterprise Edition, including Enterprise JavaBeans. While Enterprise JavaBeans technology is young, it has already been widely adopted by application server, E-business system, and framework vendors.
Distributed computing has a component-object development approach. Cross-system component portability isn't yet a reality, but this approach provides a tremendous amount of flexibility in a world where time-to-market is the top criterion.
Companies need to bring redundancy and distributed computing into play to create scalable solutions--unfortunately, this is just a small piece of what's needed. A Web site needs to be hosted in a location that can provide fast, redundant Internet access lines. Just to be safe, this site should be replicated in a different location so that in the event that one server fails, the other can continue handling traffic. Once this has been taken care of, load balancing can be put into action. Products such as Cisco Systems' Local Director and Alteon WebSystems' Web switches distribute loads across servers.
As an example, GTE Internet, a division of GTE that focuses on Internet access and hosting services, supports more than 500,000 customers and handles more than 15,000 connections per second on its mail system. "With Alteon's Web switches, we've been able to improve the availability and throughput of our systems, giving users a much better online experience," says Jay Hickman, manager of online security operations for GTE Internet. "Previously, we could only run our systems as fast as our firewalls could process traffic. With Web switches, we can better control and distribute traffic among multiple active firewalls."
However, load balancing alone isn't enough to ensure Web application availability. It can guarantee that requests are routed to the server being used the least, but it can't provide fault tolerance for complex Web applications.
Illustration by John Howard
Back to This Week's Issue
Send Us Your Feedback
Top of the Page
Featured Microsite
