The IT World article caught a very interesting definition of availability Oracle used. Oracle pledges 99.5% availability but allows itself to remove some types of outages it its availability calculations. For instance, from Oracle’s document, the following events do not count as downtime with respect to their availability pledge:
“Unavailability of management, auxiliary or administration services, including administration tools, reporting services, utilities, or other services supporting core transaction processing”
I am not quite sure how to interpret this. It seems that any number of Oracle’s internal management tools can go offline and and have some undefined impact on customer’s applications. Presumably these failures would only interrupt management activities. In that case it makes sense not to count the time as unavailable. However, the business impact of these management problems on the customer is not clear.
“Outages occurring as a result of any actions or omissions taken by Oracle at the request or direction of Customer”
This seems to be a big one. If the customer requests help from Oracle, any outages as a result of that request are not measured against their 99.5% availability pledge. I do not understand why they would not be.
“Outages due to system administration, commands, or file transfers performed by Customer users or representatives”
Surely Oracle cannot be held responsible for a customer bringing down his own applications. But is it possible for a file transfer to do so? I am not sure what that last phrase is implying.
Any time a service provider documents their SLA the nerdy side of me enjoys read it. The service provider market is evolving rapidly. And while some of Oracle’s statements may be vague I believe they have been more bold than Amazon in documenting their intentions. As these documents become more and more clear it will be easier for customers to choose among providers.