|99.4%!?!?! The SLA is 99.5%!!!|
From my experience the SLA is looked at for one thing only, the uptime. The availability, the 99 and how many more 9s will they provide? Then whenever there is an unplanned hiccup in the service you pull out the contract and swing the SLA around demanding everything under the sun in return. The ultimate finger pointer.
Here's the irony of it. Yes, contracts are contracts and they need to be upheld, supported, and honored. But from my experience the SLA numbers being held as concrete leverage of some kind is laughable. Unless you were paying a fleet of people to support your double datacenter HA configuration that costs far beyond the Cloud Service, I challenge you that you weren't meeting that SLA metric. When you had unexpected outages with your on-premise implementation you weren't screaming at your engineers demanding payback of salary for a 93% SLA instead of the 100% your upper management promised some other high level manager. Did you?
No matter how automated and expansive the cloud service is, it's still built, maintained, and touched by humans. Rumors out there is that Google has close to a million servers driving them all over the world and they still have large scale, unexpected outages. Twitter, Amazon, Microsoft, big and small from time to time don't make their advertised SLA target.
Hiccups happen but I am confident and very comfortable that the frequency and duration of any disruptions will be far less impactful than on-premise implementations. Then again there is a limit to the tolerance and if you are dealing with the vendor that can't keep the lights on for you to do business, then you have a legitimate case to demand action. When it comes to the infrequent disruptions, calm down.
End of Line.
Twitter - @BinaryBlogger
Facebook - Binary Blogger