Google App Engine Cloud Availability Plummets, Sparks Outages

July 16, 2011 Off By Hoofer
Grazed from CRN.  Author: Andrew R. Hickey.

Google (NSDQ:GOOG) App Engine, the search giant’s cloud development platform, suffered a massive performance hiccup on Thursday when availability dropped and knocked some user Web sites off line.

According to Google’s App Engine status dashboard, the cloud platform which developers use to run Web and cloud applications written in Python or Java, availability nosedived between 9 p.m. and 11 p.m. PST Thursday, taking down some customer apps and Web sites built on App Engine. According to a report from CNET (NSDQ:CNET), availability fell to at least 88.1 percent.

According to the Google App Engine health status dashboard, the cloud development platform suffered issues serving Java programs, which created errors and problems with requests to an app’s Web page resource. There were also delays in using the programming interface to manage tasks.

The dashboard indicated that Python applications weren’t affected and that the Java application performance issues were righted shortly before midnight.

On its App Engine status dashboard, Google said App Engine has been fixed, and the company is probing the root cause.

"This issue is resolved, but we are currently conducting an investigation," Google said. "We will follow up with more details."

Google’s dashboard also indicates that the company was performing scheduled maintenance on Google App Engine between 6:04 p.m. to 11:59 p.m. PST Thursday.

"We had a scheduled maintenance period today where the datastore was put into read-only mode and memcache flushed," Google said, explaining the massive spike in latency that occurred around 11 p.m. PST on Thursday. It was unclear if that maintenance and the performance issues were related.

Google did not respond to requests for additional comment.

Some Google App Engine users were angered by the lack of availability.

One App Engine developer and maker of the online word game Wordament wrote that the service blip knocked the game off-line for as many as four hours. Wordament is the first game by Seattle area mobile and Web game studio You vs. the Internet.

"Wordament has been down for 4 hours now due to a continued outage of Google App Engine (our infrastructure provider)," Wordament wrote on its Web site early Friday morning. "We are aware of the issue and doing what we can to return service to life ASAP. We are really sorry to not be able to play with you. More updates as we learn more…"

In an update more than two hours later, the Wordament team said it was back in business.

"Well, I think I aged a few years tonight. That really, really sucked. The worst part? Google never acknowledged that they had a problem," a Wordament developer wrote. "Down the line of their Service Status was ‘Normal.’ And yet, we were running 0 servers. The other part that really sucked was our powerlessness to do anything. We couldn’t even contact support. Sigh."

The Wordament team said the issue opened its eyes to issues that can arise in the cloud.

"Believe us: we will be considering our options for redundancy," Wordament wrote. "It seems that one cloud may not be sufficient. Thanks for your understanding tonight, but I do expect a bunch of bad reviews from new players that tried to sign up and got the horrible experience."

Google App Engine joins a growing list of cloud computing players suffering availability issues and cloud outages this year. Most notably, Amazon Web Services’ cloud offerings suffered a massive outage in April which knocked out a host of Amazon (NSDQ:AMZN) cloud customers’ Web sites in some cases for several days.

The Google App Engine hiccups also follow Google’s decision to kill Google App Engine for Business and wrap the business features into the standard edition of Google App Engine.