EBay: One outage too many
|
|
|||
|
|
Sign up to receive this and other networking newsletters in your inbox.
When it comes to outsourcing, there are two mistakes a company can make. The first is to overuse outsourcing services and become inextricably dependent on a third-party and the second is for a company to blindly ignore the option of outsourcing, even when it becomes obvious that the company needs some outside help.
EBay, the popular online auction site, might be the poster child for this latter group of mistake-makers.
On April 7, eBay suffered an outage of more than nine hours on its Web site when its popular clothing and jewelry auctions were rendered inaccessible to both buyers and sellers. The company did not reveal the cause of the outage, but it wasn't the auction site's first this year - just last month, eBay lost its home page for more than 90 minutes, and then experienced performance slowdowns again a few weeks later.
Availability problems at eBay have become almost an annual event. In January of 2001, the site crashed for 11 hours. In June of 1999, eBay was unavailable for 22 hours, the culmination of some 15 outages that occurred between August 1998 and November 1999. And these are only the major outages - any eBay regular will tell you that the site has experienced innumerable smaller glitches with its search engine and its auction upload systems over the past five years.
To its credit, eBay has attempted to fix its availability problems via internal IT efforts. Since 1998, the company has changed CIOs, installed redundant server back-up capabilities, and initiated a new application framework based on Java 2 Enterprise Edition and Enterprise JavaBeans. The company has distributed its auction functionality so that a single server failure will only affect some parts of the site, rather than causing sitewide blackouts such as those that happened in 1999.
Despite all of these efforts, however, site availability continues to be a troublesome - and highly visible - problem for eBay. Nearly five years after the first outage, small businesses that rely on the auction provider as a primary sales channel are still complaining of lost customers and lost sales as a result of the site's technical glitches.
At some companies, that sort of chronic availability problem might cause top brass to change IT horses, either bringing in new staff or calling in an outsourcing vendor. EBay has added some hosting services to its IT mix, but the company has steadfastly refused to do any major outsourcing, citing a need to keep control over its unique - and industry-dominant - auction technology.
After five years of outages, however, an outsourcing vendor is exactly what eBay needs. A good systems integrator - or even a knowledgeable consultant - could help the company to assess its site architecture from a broad perspective, objectively pointing out the mistakes that the company has made over and over. An architectural review might also help eBay to identify future business directions that might impact performance, and build capacity to handle the load.
A more strategic view on outsourcing might also help eBay manage outages when they do occur. A strong hosting provider might offer problem management services as well as server and network backup in an emergency. Many vendors also offer external monitoring services that can alert companies to problems that may affect end-user experience on a corporate Web site.
EBay's staunch desire to maintain its own operations is admirable and - given its auction industry dominance - understandable. But after five years of major outages, it is time for eBay to admit that it needs some help. When you've been staring at the same puzzle for too long, sometimes it takes a fresh pair of eyes to find the missing piece.
RELATED LINKS
IDG News Service, 04/03/02
Senior Analyst Tim Wilson is with Enterprise Management Associates in Boulder, Colo., an analyst and market research firm focusing exclusively on all aspects of enterprise management. Wilson has over 10 years of experience in covering e-business and enterprise management issues, most recently with InternetWeek, where he was chief of reporters. He can be reached by clicking here.
