The Outlook Outage Explained: What Microsoft Broke - by a Retired Microsoft Engineer.

My wife added a new mailbox to Outlook mobile just before failures began so, pretty sure that was the cause of failure. Sorry...

Replace Satya with a MS greybeard. This needs to stop now.

If they would have downloaded TMOG the whole world would have been back up sooner. Just sayin'.

Do the FAA and Amtrak next.

my email worked (Google) my personal domains worked (cloudflare for dns, and hosting at a midsize hosting firm) "All our eggs in one basket" is a bad strategy for any business hosting solely with microsoft (or any cloud, for that matter).

going to venture a guess...

Microsoft hasn't released the technical breakdown so how can you know ?

Watch the video and find out!

Still waiting on my t-shirt with your face on it though

Microsoft364 lol

@crimecontainer remember when I said X has more experts per capita, this is why

its aliens, we already know

@grok can you summarize the video?

Fantastic breakdown. I tried to explain this (in a nutshell) to a pissed off cop this afternoon who couldn’t quite understand why he couldn’t get authentication emails before they timed out. He feigned rolling his eyes in the back of his head but he at least stopped yapping.

That's what i wanted to say yes Outlook is just the Client basically. In our Company we mostly use Exchange Online but some Mailboxes are still on the Exchange on-prem servers.

Microsoft will soon fall to linux, the world will retire it like a covid plauge.

When AT&T had 4ESS handling catastrophic loads from phone-ins and 555 numbers from TV being dialled opportunistically, they invented call-gapping. This was roughly the same as cisco IOS: Weighted Random Early Detection for queue management, but the connection oriented equivalent. Varnish and Nginx are built this way, in Web session parlance. There should be no element of auth, DNS, distributed storage, queueing theory in action, or congested network link that should collapse a state machine or threaded dispatcher equivalent. Historically, if you bent IIS with too much web traffic, or ran out of ephemeral ports on the network stack, the Linux equivalent popular suite that did the same job would have an edge. But, we're well past boasting eras for most parts of modern compute. But, patches and on-the-fly patches can throw up old friends that used to bomb out our older tech/compute. My concern is that cloud Exchange, ie Online is generally perceived to be ahead of on-premise Exchange. And presumably, the clustered implementation with its opportunistic-locks and MS-SQL-like storage paradigms brings a gulp moment if mixing the stprage environment with two concurrent machines sharing the same user data across different releases or different tracks of the software that is causing perceived operational issues in the current incarnation. Indeed, Have, we will have to wait and see. Incidentally, the 4ESS switch gapping was marvellous and had similar approaches put into data network solutions such as with Digital Communications Associates, DCA, with their multiprotocol X.25, LAT, Telnet, DDCMP, and DEC Unibus handout technology. Gapping calls for bulletin-board menu 'node hosts' with routing functions from the Z80 Node Supervisor derived WAN path and resource allocation logic that coukd serve many thousands of users all trying to contend with a few hundred interactive terminal/PAD sessions on any given menu to a host provider. Sometimes the old lessons and protection paradigms best describe modern needs and how to solve them whilst remaining productive rather than being too negatively influencing. —Realising that with AT&T, the phone handsets and 5ESS and PBXs had limited verb sets compared to Outlook, Teams, and Copilot apps/portals/APIs. However, if the line seizure rate dipped or call concurrency reached a maximum termination status for circuit-switched time-space-time functions, the gapping technique optimised getting service resources into a fully occupied state to a practicable level that brought satisfaction. What we saw today was a chaotic swing of badness, with prior knowledge in the art of queueing, thoroughly driven into the ground. It would seem that the data access modes of Exchange should solve resource issues of any element so it cannot cascade a failure condition to bring down the whole caboose, and perhaps a stateless interlock to messages in-process that implicitly allow any new node to continue on with a partially completed job. Which may be the case, but confidence that kicking in new resource working in anyway well, seems lacking. Most human ventures lead to learning from mistakes. Let us hope that the Bell Labs professors and their innate inventiveness has not been lost. Every generation has amazing specialists and architectural geniuses. We have to believe in that! As a final aside, in the UK wing of globalised AT&T, in the 1990s we had a memo circulated ftom the top-brass derived ftom an employee suggestion-box, the program was called IDEAGEN, a portmanteau of Idea and Generate. In this memo it was stipulated that emails should be read, but left unanswered until later. Largely, all left in the mailbox (RSTS PDP-11 COMET email) until the afternoon. For a delivery/fulfilment organisation, it initially helped, but it ultimately died out for the next bright idea. We basically did the Microsoft outage equivalent to encourage productivity. Then we added no internal phone calls in the morning too!

No better person, could’ve explained it👨🏾‍💻

Opining on current issues when you peaked in your Novell days is a wild flex brother.

I want the last ten minutes of my life back

Microsoft has so many authentication layers its like patchwork bound to happen eventually

Funny how their Azure(cloud) proved to be worthless.

Days before 365 started misbehaving with dmarc and spf

@Grok What's he trying to say here in a nut shell please? This guy rambles too much

✈️📡

Can you explain why @azure is still down globally.. If the FAA admits their unvetted $875 million Silicon Valley AI toy choked on a degraded cloud backbone and saturated the ground gateways, heads roll and commercial cloud vendors face billions in cyber insurance liability....

I do not use MS for mail services and my email was fine, I could email back and forth with iCloud for example. What I found interesting is I was still receiving emails from Amazon (?) and some lesser-known businesses.

Someone ask Copilot for help and got rekt is what happened. When all services started to resume, copilot stayed offline.

They need you back there Dave to sort out the crap.

It’s AI slop guy.

A metwork update? Seems like a stretch.

Outstanding update. Thank you!!

I remember when try tried to roll out windows 2k, they DNS requests using tcp instead of udp. It flooded their own servers down. Prior to that, Freebsd was the front end and Solaris on the back end, storage. Using pre-beta Sun storage arrays, with no documentation