November 14, 2012, 2:35 PM — Microsoft's Office 365 service has suffered two email outages within a week of each other that affected some customers in North and South America that stemmed from different causes but ended in the same result: failed email delivery.
The first outage Nov. 8 stemmed from an overwhelmed antivirus engine and the subsequent backup that caused the service degradation. The second on Nov. 13 resulted from the failure of unspecified network elements, routine maintenance and increased load that combined to degrade service, according to the Office 365 blog posted by Rajesh Jha, the corporate vice president of Microsoft's Office division.
He didn't say how many customers were affected or where they were located other than somewhere on the two continents. Both outages affected just Office 365 Exchange Online mail services.
Affected customers are entitled to a service credit. Jha apologizes and promises a post mortem on the outages as well as an update on how the Office 365 service level agreement was affected.
The Nov. 8 incident started when an antivirus engine bogged down as it processed emails that the engine determined carried a particular virus. That delay processing emails led to retries that further bottlenecked email flow including legitimate emails, he says.
The issue was resolved by intercepting the tainted messages and quarantining them directly.
To head off similar problems down the line, the company has set a lower threshold for diverting problem emails and implementing faster remediation tools. It is also adding unspecified safeguards that automate remediation of this type of problem, Jha says.
The second incident Nov. 13 started with some scheduled maintenance that required shifting some of the load out of those data centers undergoing maintenance. During this work unspecified network elements failed but sent no alerts of their failure, he says. And finally the entire infrastructure was handling more traffic from new customers, all of which resulted in some customers being unable to access email services.
Traffic for affected users was shifted to healthy data centers while the issues were dealt with.
Jha says the company is in the midst of increasing capacity and is automating how equipment failures are handled to speed up recovery time.