The outage occurred Tuesday morning and again in the afternoon before Google engineers were able to isolate the problem and bring the system back to normalcy. The Google App Engine enables developers to build applications and run them on Google's infrastructure. The service is available at no charge for applications using up to 500 MB of storage and enough bandwidth and CPU power for 5 million monthly page views.
In a developer discussion forum, a Google engineer who went by the name of Pete confirmed that the outage occurred in a data store about 6:30 a.m. Pacific time Tuesday, causing error returns to a "small percentage of requests."
The number of error returns increased "significantly" between 9 a.m. and 11:30 a.m. and again at 12:40 p.m. before Google was able to isolate the problem and bring the system back to normal, the engineer said.
Google was not immediately available for comment Wednesday, and details were scarce. But Pete said the outage was the result of a flaw in Google's data-store servers and was trigged by a particular class of queries.
"We have isolated the bug and we're currently working on a fix," the engineer said. "Going forward, we're also working to further isolate queries so that in the future a bug like this won't affect the stability of the system as a whole."
Other Internet companies, such as online retailer Amazon.com, are also offering so-called "cloud computing" services. Potential corporate customers, however, have remained wary, because of a number of concerns. Namely, would they get locked into a single vendor with their data or applications held hostage, or would consumer-oriented vendors such as Google and Amazon stay in the corporate IT business for the long haul? There's also the question of government regulations that may not allow certain data to be hosted outside an organization.