There is a difference in how individual and total outages are reported by the SLA page within Azure. It is calculating the end of the individual outage using the start time of the last bin instead of the end time (see red line on the image below). This means that the outage time is rounded down (see success values circled in purple). The total outage time (see green line on the image) counts the whole of that last bin so there is an obvious inconsistency. Microsoft is currently saying that the feature is not widely used so they won't fix it. Does the bug affect anyone else? If so could you post your support for it being fixed please?