The cause seems unclear, and there are a few competing theories going around. What should we look for in the postmortem?
Do we know what caused the Microsoft outage yet?
Answers
I would wait for the root cause analysis before settling on an explanation. A slow or changing investigation update doesn't necessarily mean the engineers have no idea what's happening. They may still be separating the initial fault from the things that broke afterward.
The useful parts of the postmortem will be the trigger, why it affected so many services, how they restored service, and what changes are supposed to stop a repeat.
Reply to this answer
DNS, Front Door, virtual network configuration and BGP have all been floated as possibilities. That is a pretty broad list. A global outage alone doesn't tell us which layer failed, so I wouldn't treat any of those guesses as a diagnosis.
Reply to this answer
The support side worries me as much as the technical cause. When an outage drags on, customers need useful updates and access to people who can investigate. Staffing and lost expertise have been raised as concerns, but there's not enough here to say they caused this outage.
Reply to this answer
There is also speculation about whether this was an internal failure rather than an attack. Either way, that needs evidence. A postmortem should explain what was found rather than asking people to infer the cause from the outage itself.
Reply to this answer
Add your answer
Answers are reviewed before publication.
8c!fablepocket_392:
So even if one of those services was affected, it might not have been where the problem started?
_b82+quendavo.6:
Right. A service can show symptoms because something it depends on is broken. The timeline and dependency details should help distinguish the two.