Salesforce.com outage darkens cloud computing

Salesforce.com suffered a major outage earlier this week, leaving thousands of its customers unable to access applications.

The downtime affected the software-as-a-service (SaaS) vendor's data centres, which it uses to run its cloud computing systems. It affected users for almost an hour on Tuesday.

The vendor directed any request for comment on the cause of the outage to its trust.salesforce.com community page, which said a core network device failed at 8.39pm GMT preventing the processing of any data.

When the automatic failover to backup systems also failed, attempts were made to manually restore the service, which was up and running by 11.00pm that evening.

Afterwards, Salesforce.com stated that it was confident the root cause of the outage had been addressed by the manual workaround.

But it added: "The Salesforce.com technology team will continue to work with hardware vendors to fully detail the root cause and identify if further patching or fixes will be needed."

Although the outage occurred outside UK business hours, some US customers using Twitter reported that the outage only affected them for short periods of about five minutes at a time.

Nevertheless, the outage may add weight to those who have questioned the resilience of SaaS-based enterprise IT delivery.

Despite Tuesday's hiccup, Quocirca service director Clive Longbottom told IT PRO that Salesforce.com had a relatively good track record on service uptime. "It's not as bad as some of the outages we've had from the likes of RIM, where the service was down for up to a day," said the analyst.

"And you still get good levels of service compared to those delivered from within an organisation that is backed with solid service level agreements."

"Overall, Salesforce.com is pretty good, but I wouldn't say pretty good was good enough," he added.

Longbottom said only an external network problem, like the recent internet disruptions caused by damage to undersea cables, should be a sufficient excuse for SaaS downtime.