Much has been made of the new wave of web services, where a hosted application fulfills the needs of numerous users. When it goes wrong, though, the efficiency of the distributed system plummets.
Client/server technology isn’t a new idea. It has been around for decades, and the model has seen a resurrection under the label of Web 2.0. Now everyone calls it web services.
They’re great when they are available. Anywhere the user goes, the application is sure to follow. When they go away, as Salesforce.com did for its customers for several hours recently, the reason why client/server comes and goes becomes apparent.
18,700 customers of the hosted CRM service found themselves without application availability for over eight hours. Salesforce.com blamed a faulty database for the issue while repairing the service. A failover that is supposed to take place when this happens to keep the service running didn’t happen.
In the report from Silicon.com, Salesforce apologized for the problem, and noted they have greater than 99 percent uptime. Some customers cited in the article disagreed, claiming monthly outages of an hour or longer occur regularly.
The problem comes down to infrastructure. A business has to have multiple instances of its web service available, lots of redundancy, to be able to stand behind claims of reliability.
Without that redundancy, web services can and do go away, especially as more users crowd into the system. Six Apart had that problem recently with its Typepad blogging service, a problem that persisted for months.
Web services may offer alternatives to PC-based applications, but until they reach a point of reliability greater than 99 percent, Microsoft doesn’t have any reason to worry about its desktop businesses, especially Office.
—
Email the author here.
Add to document.write(“Del.icio.us”) | Yahoo My Web
David Utter is a staff writer for murdok covering technology and business.
