Performance Vs. Uptime

There's a practical difference between maximizing uptime and improving performance. Ensuring uptime calls for redundant bandwidth, routers and Web servers. Enhancing performance requires the addition of load balancers and caching services to give users the shortest path to the content they need.

Today, either uptime is increasingly viewed as a given or a minimal amount of downtime is viewed as acceptable.

"If a site is always available but slow, a company is not achieving its objectives from a customer standpoint," says Jeff Banker, vice president of Gomez Networks, one of the many firms that monitor site performance. "If a site is fast yet unavailable infrequently, it's still meeting expectations." Banker points to eBay--which has experienced infrequent but prolonged outages--to support his assertion that users will stick with sites they find valuable.

Indeed, companies lose more than $4 billion a year in abandoned transactions because of slow Web sites, Zona Research estimates.

This new performance push helps explain the proliferation of monitoring services from the likes of Exodus Communications, Freshwater Software, Gomez, Keynote Systems, Mercury Interactive, Service Metrics and newcomer WebHancer. They analyze how sites are performing from the perspective of their customers. In some cases, they recommend adjustments to improve visitors' experiences.

"Availability is assumed at this point, like the dial tone on a phone," Banker says. "But performance is often lousy, and that affects business."

Monitoring can help companies identify problems occurring outside the corporate infrastructure. Online stock broker Ameritrade used test data from Keynote to pinpoint and correct problems emanating from an ISP, says CIO Jim Ditmore.

Such problems sometimes stem from faulty connections between ISPs. In Ameritrade's case, one ISP boasted T3 capacity at its end points, but that bandwidth was throttled back to T1 on the parts of the network the ISP doesn't own.

"That's like having two fire hoses connected with a garden hose," Ditmore says.

Since going to its ISPs armed with Keynote data demonstrating the performance degradation this problem caused, Ameritrade has has been able to work with providers to ensure more consistent bandwidth and more favorable peering. The result: 30 percent better bandwidth utilization.

Another financial services provider, Deutsche Bank, has used Mercury Interactive to identify and correct performance problems. Bank assistant vice president Mark Neuwirth notes that "good site performance is my day-to-day goal."

Although Neuwirth had in-house network engineers checking every element of the network from the firewall to the Unix servers, he found that content partners that deliver information on the Deutsche Bank site were degrading performance. Specifically, the bank has partnerships with research firms that provide feedback surveys from their own servers.

Server problems among such partners were causing the Deutsche Bank site to time out, preventing Deutsche Bank site visitors from making or maintaining connections.

Response time is especially important today for United Parcel Service, since customers now track UPS deliveries in progress from their cell phones and their own Web sites, says Marc Dodge, manager of Internet and intranet systems.

"We don't have downtime, but if we have performance problems, we get a lot of heat," Dodge says.

User tolerance for slow performance varies by site and the importance of the information they're seeking, says Bruce Linton, president of performance monitor WebHancer.

"People who visit the World Wrestling Federation site are so loyal," Linton says, "they will sometimes wait 80 seconds for a page to download. But with Yahoo, expectations are much different. People are much less forgiving."

CNN Interactive knows it has little margin for error; its visitors can go to other news sites to get the same information.

"We can't afford to struggle; not if the competition is keeping up," says Monty Mullig, executive vice president of the technology group at CNN Interactive. "Our job is staying alive longer than the next guy."

While uptime is an easy variable to measure in-house, performance measurements require analyzing factors over which an e-business has little control, such as Internet performance, which can vary greatly because of a lack of uniform quality-of-service standards across the Internet. That's why CNN Interactive uses Keynote and WebHancer, plus homegrown monitoring tools. Based on these various data sources, the company has added redundancy as well as capacity for major news events The steps that these big organizations are taking to boost performance are clearly paying off. Three years ago, Keynote started tracking the performance of the top 40 business sites. At that time, their pages were downloading at an average of 7.8 seconds. Today, the average is 2.5 seconds.