Microsoft's Exchange cloud stumbles

Poor outage disclosure leads to policy change.

Microsoft has apologised again to business customers for three outages that hit clients running Exchange on its Business Productivity Online Suite (BPOS) cloud. 

The first incident on Tuesday affected clients on Microsoft's American network for between six to nine hours, with little explanation from the vendor beyond the outage status on its service dashboard.  

One customer complained at the time the dashboard passed the event off as a “service degradation” when it was actually an “interruption”. 

Two days after the initial outage Dave Thompson, Microsoft’s corporate vice president for online services offered an apology and explained that hub components that support its BPOS-S Exchange service did not work correctly after being hit by “malformed email traffic”.

“I’d like to apologise to you, our customers and partners, for the obvious inconveniences these issues caused,” said Thompson. "We know that email is a critical part of your business communication."

Microsoft’s network then suffered two more hiccups caused by malformed traffic and another unrelated problem. These incidents caused a backlog of 1.5 million emails. 

Some customers were also unable to access Outlook Web Access, which Microsoft discovered was triggered by problems with its servers hosting the mail.microsoftonline.com domain.

Customers took to Microsoft’s support forum to express outrage at the continued disruptions and lack of information, including one who claimed that since switching from in-house Exchange to BPOS that it now paid more but had less reliability. 

Thompson conceded Microsoft could have been more open with customers from the outset, and has since changed its communications policy for the service. 

“As a result of Tuesday’s incident, we feel we could have communicated earlier and been more specific.  Effective today, we updated our communications procedures to be more extensive and timely,” said Thompson. 

Amazon Web Services made the same concession after its recent outage.

Thompson also promised a full post-mortem of the event.  

Microsoft apologised for BPOS in September last year after several outages, which it later admitted were caused by a network upgrade.