An outage at software-as-a-service pioneer Salesforce.com yesterday has provoked an angry reaction from some customers, and raised further questions about the reliability of the on-demand application delivery model.

Salesforce said that the service disruption affected all areas from 20.39 to 21.17 GMT on 6 January. "A core network device failed due to memory allocation errors," reads the incident report.

It is unlikely, therefore, to have affected many UK-based workers, but has caused deep concern among some US customers.

Many turned to their Twitter community to find out more details about the outage, as the vendor's trust.salesforce.com system status site was also down.

"Salesforce. Don't fail me when I need you the most!" wrote one Twitter user, while another said: "Salesforce crashes, Twitter/Google get hacked, what's next? I am going to switch to landlines and snail mail?"

Another user said: "Salesforce was down again today: ammunition for the on premise argument."

However, others were more forgiving. "I adore Salesforce.com; been using it for years. It has rarely hiccupped, which makes today remarkable (but not scary)," said one.

Another argued: "In a past life, a server outage would have been my problem. Now I let Salesforce do the dirty work."

Rob Bryant, a partner at consultancy Deloitte, argued that few companies would be able to achieve similar resilience to Salesforce.com in a cost-effective way in-house.

"Salesforce.com has an excellent track record, but on the rare occasions
that things do go wrong CIOs can sleep soundly knowing the right senior
engineers are working on the problem - engineers that their firm alone
may struggle to access with the same level of urgency," he added.

However, Steve Moyle, chief technology officer of database seucity firm Secerno, said the incident raises serious questions of trust for Salesforce.com.

"Almost a million users who have out-sourced a mission critical business function are paralysed and not able to access their own customer data, weakening their trust in the system," he added.

"Data owners must ensure that all of their data is fully protected and controlled to the point that no inappropriate use of the data can occur."

Salesforce.com said in a statement that it takes service performance "very seriously", and that any further information about the outage would be displayed on trust.salesforce.com as it becomes available.