We just keep going and going and going

Interesting post from on the cloud downtime dilemma from Software Advice. It cites a Radicati Group study, which found that among popular on-premise email systems, there was an average of 30-60 minutes of unscheduled downtime per month, in addition to the 36-90 minutes of scheduled downtime – so between 66-120 minutes of total downtime per month. Google claims an average of 7 minutes of downtime per month during 2010 for Gmail. Of course, it’s had some known issues in 2011.

But cloud solutions tend to be more transparent than their on-premise counterparts in general. So when something happens – whether crash or breach - news of the problem is disseminated faster than it would be if the problem occurred on-prem.

From our beginning, we have structured our SLAs so that there’s no provision for scheduled downtime. That means even during our weekly product upgrades, we never take down your instance of Okta. Of course, when you’re in the business we’re in – where customers rely on the user authentication component of our service to access their business critical applications – it’s an absolute must.

We embrace full transparency at Okta. We post raw data of our downtime every month at www.okta.com/trust/.

Note that even during the month of April – with the fallout from Amazon EC2 – we were down for 10 minutes. This is what we mean by high availability – we will just keep going...